Radio receiving, recording and playback system

Radio receiving, recording and playback system
US 20030163823A1
(19) United States
(12) Patent Application Publication (10) Pub. No.: US 2003/0163823 A1
(43) Pub. Date:
Logan et al.
Aug. 28, 2003
Continuation-in-part of application No. 10/165,587,
?led on Jun. 8, 2002.
(75) Inventors: James D. Logan, Candia, NH (US);
Daniel M. Morton, Sornerville, MA
(60) Provisional application No. 60/346,418, ?led on Dec.
29, 2001.
Publication Classi?cation
Correspondence Address:
Int. Cl.7 ........................... .. H04N 7/173; H04N 7/16
WEST YARMOUTH, MA 02673-2516 (US)
US. Cl. .......................... .. 725/89; 725/134; 725/142;
(73) Assignee: Gotuit Media, Inc.
(21) Appl. No.:
Dec_ 30, 2002
Related US, Application Data
A radio receiver and storage unit Which concurrently and
continuously receives and records a plurality of separate,
simultaneously broadcast programs, and then selectively
(63) Continuation-in-part of application No. 09/238,948,
buttons selectively perform the following ?ve functions not
reproduces desired programs at desired times. User interface
?led on Jan. 27, 1999.
available on a conventional radio: Pause: suspends the
Continuation-in-part of application No. 09/536,969,
reproduction of the live or recorded programming currently
?led on Mar. 28, 2000.
being played; Save: permits the listener to save the complete
Continuation-in-part of application No. 09/699,176,
content of a live program currently being reproduced; Juke
?led on Oct. 28, 2000.
Continuation-in-part of application No. 09/782,546,
box: permits the listener can select and playback previously
recorded programming; Mark: alloWs the user to “book
?led on Feb. 13, 2001.
mar ” a speci?c position on a program; and Options:
Continuation-in-part of application No. 10/060,001,
?led on Jan. 29, 2002.
permits the user to obtain information about available pro
gramming, or to perform less frequently used functions.
‘F —————————————————————————————————————5 El
21 1
D t
e 90 or * Decoder
230 241
coder 7"
£53 1
l_ _ g ‘ g _ _ _ i g ? _ _ i
i _ * ; ________________ _ ‘l
2nd Tuner
Bandpass _
Third Tuner
Fourth Tuner
Patent Application Publication Aug. 28, 2003 Sheet 1 0f 2
US 2003/0163823 Al
Auto Battery
Audio Out —>
256 MB
K i
(Fig. 2)
_r\___ >
Audio Out-b
LCD Display
<—o 0 o o 0
CD Player
: /-117
Aug. 28, 2003
US 2003/0163823 A1
[0001] This is a continuation in part of, and claims the
bene?t of the effective ?ling date of, each of the following
pending US. patent applications, the disclosures of Which
are hereby incorporated herein by reference: U.S. applica
tion Ser. No. 09/238,948 ?led on Jan. 27, 1999 by James D.
Logan entitled “Apparatus and Methods for Broadcast
[0008] FIG. 2 is a schematic block diagram shoWing the
details of the parallel tuner/encoder mechanism used to
simultaneously receive and digitiZe the program content
broadcast by several different selected radio stations.
[0009] In the folloWing description, an overvieW of the
hardWare employed to implement the invention Will be
presented ?rst, folloWed by a more detailed discussion of
speci?c features and functions.
Monitoring and for Providing Individual Programming;”
[0010] HardWare OvervieW
US. application Ser. No. 09/536,969 ?led on Mar. 28, 2000
by James D. Logan et al. entitled “Systems and Methods for
[0011] A car radio receiving, recording and reproduction
Modifying Broadcast Programming;” US. application Ser.
No. 09/699,176 ?led on Oct. 28, 2000 by James D. Logan
et al. entitled “Advertising Supported Music Delivery Sys
tem;” US. application Ser. No. 09/782,546 ?led on Feb. 13,
2001 by James D. Logan et al. entitled “Broadcast Program
and Advertising Distribution System;” US. application Ser.
system Which embodies the invention is shoWn in FIG. 1.
Although the speci?c embodiment to be described receives
radio signals, and records and reproduces the audio program
content of those signals, it should be understood that, in the
main, the principles of the invention may also be applied to
the reception, storage and playback of television program
ming as Well.
No. 10/060,001 ?led on Jan. 29, 2002 noW published as Pub.
No. 2002-020925 A entitled “Audio and Video Program
Recording, Editing and Playback Systems Using Metadata;”
and US. application Ser. No. 10/165,587 ?led on Jun. 8,
2002 by James D. Logan et al. entitled “Audio and Video
[0012] The system includes a plurality of separate tuners
and connected encoders, shoWn collectively at 101 and in
detail in FIG. 2. Each tuner/encoder combination is capable
of selecting a received radio signal and converting that
Program Recording, Editing and Playback Systems Using
signal into digital form. The multiplexed digitiZed content
from the tuners and encoders 101 is passed to digital signal
processor (DSP) 103 Which compresses the digital content
of each signal and stores the compressed signal in a high
[0002] This application further claims the bene?t of the
?ling date of US. Provisional Patent Application Serial No.
60/346,418 ?led on Dec. 29, 2001 by James D. Logan et al.
entitled “Radio Receiving, Recording and Playback Sys
capacity (eg 256 MB) random access memory (RAM) 105.
The DSP executes a variety of programs, to be described,
Which are stored in a program memory (not shoWn).
This invention relates to methods and apparatus for
receiving, recording and reproducing radio and television
signals and more particularly, although in its broader aspects
not exclusively, to a radio receiver Which incorporates
means for concurrently recording several live broadcasts
Which may be selectively replayed at a later time.
It is a leading object of the present invention to
permit a radio listener or a television vieWer to enjoy the
programming they Want to Watch or listen to When it is most
[0005] The present invention operates by the concurrently
and continuously receiving and recording a plurality of
separate, simultaneously broadcast programs, and then
selectively reproducing desired programs and desired times.
[0006] These and other objects, features and advantages of
the present invention may be made more apparent by
considering the folloWing detailed description of a speci?c
embodiment of the invention. In the course of this descrip
tion, frequent reference Will be made to the attached draW
FIG. 1 is a schematic block diagram of a radio
receiving system embodying the invention; and
Program content selected by the user may also be
read from or persistently stored in a Writable ?ash memory
107, or on a compact disk using the CD player/recorder 109.
The parallel tuners 101 and the DSP 103 operate under the
control of a user interface control 111 that responds to
control inputs from a set of pushbuttons 113 manipulated by
the listener, and that operates a visual liquid crystal display
(LCD) 115 for displaying information to the listener. An
external data port 117, Which may take the form of a
universal system bus (USB) connection or a netWork port
(eg an Ethernet connection), may used to exchange both
digitiZed programming content and metadata that describes
available programming content With a connected computer
or device.
[0014] The system is preferably poWered by the automo
bile battery 121 through a direct connection 123 to permit
incoming radio signals to be recorded even When the car’s
ignition sWitch 124 is off, While a second connection 125
through the ignition sWitch 124 poWers selected functions
only When the car is in use. A small additional battery, as
used in rechargeable cell-phones or laptop computers, may
be employed in combination With a charger that operates
When he engine is running.
Tuners and Encoders
[0016] The details of the parallel tuner unit 101 are shoWn
in FIG. 2. The output signals Within a desired passband
selected by a bandpass ?lter 201 from an antenna 205 are fed
to four separate, parallel, concurrently-operating tuners 211,
212, 213 and 214 Which have like components but Which are
independently tunable to receive four different stations. As
Aug. 28, 2003
US 2003/0163823 A1
discussed in more detail later, a different number of tuners
and encoders may be employed.
[0017] The components Which form the tuner 211 are
shoWn in detail in FIG. 2. An input preampli?er 221 has an
output that is fed, along With the output from a local
oscillator 223, to the inputs of a mixer 225. The loWer
frequency difference signal from the mixer 215 is passed
through a bandpass ?lter 227 Which selects the desired
station. The unit is tuned by a control signal 230 (from the
user interface 130 seen in FIG. 1) Which controls the
should be organic—With a person’s changing tastes and
perceptions about the value of What has been stored dictating
Where to draW that line. Currently, DRAM costs about 10
cents a minute or 40 cents per song. Flash upgrades could
cost four times as much. These costs should drive the
prospective siZe of the jukebox memory. Long term, small
footprint hard drives or some removable media could be
employed to increase storage capacity and portability spe
ci?cally When in use With the jukebox.
[0024] A CD player/recorder capable of playing conven
tional CD’s and further capable of recording programming
frequency of the local oscillator 223 such that the difference
betWeen the operating frequency of the local oscillator 223
and the broadcast frequency of the desired station matches
Which the listener has stored in RAM Would permit the
listener to persistently record favorite programming. The
the bandpass frequency of ?lter 227.
player/recorder 109 may thus act as an MP3 player/recorder,
[0018] The resulting signal is then passed through a detec
tor 241 and a stereo decoder 243 Which produces audio
baseband audio signals, one for each of the stereo channels.
These tWo audio signals are then digitiZed by the analog
to-digital converters 252 and 253 Whose outputs are passed
to a digital multiplexer 260 Which delivers the 8 digital
output signals (tWo stereo channels for each selected station)
as a single multiplexed stream to the digital signal processor
103 seen in FIG. 1. The reception system may be used to
receive conventional broadcast FM, AM, IBOC and other
digital transmissions, and satellite transmissions.
[0019] The number of tuners/encoders included in the
system is subject to a variety of considerations. In theory,
each listener has an ideal “skim rate”. That is, assuming a
certain type of content, a given listener might Wish to surf
through 2 hours of content in a 1/2 hour session for a skim rate
of four to one. If the listener’s skim rate is 4:1 and the unit
includes four tuners, the user achieves “equilibrium;” that is,
she can record content as quickly as she can consume it. In
general, enough tuners should be included to ensure that the
average listener does exhaust the amount of accumulated
contend While skimming.
alloWing the listener to replay music and programming using
CDs created by a personal computer, and Would provide a
convenient mechanism for offload content that has been
saved in RAM.
[0025] In addition, the unit may be provided With a ?ash
memory port 107 for accepting removable ?ash memory
cards or memory sticks, Which currently Would permit 256
MB of non-volatile memory to be added for approximately
External Data Connection
Ideally, the unit Would have some form of connec
tivity to permit data exchange With remote devices. This data
exchange capability can take place using removable media
(CD’s, memory cards or memory sticks) With the CD
recorder/player 130 or the ?ash memory port 107, or by
means of a data port such as the USB port seen at 117 Which
may be inexpensively incorporated into the unit. The data
port may also take the form of a Wired or Wireless netWork
connection, or by receiving data in the form of broadcast
radio signals. Regardless of the mechanism employed, the
external data connection may be used to:
broadcasts, and not replaying programming from the juke
[0028] a. Upgrade the softWare over time (a potential
revenue stream);
box or CD, during a given session, the demand for tuners
Note that, if the listener is tuning to live radio
and memory is reduced. To the extent the listener relies on
“scheduled” programming, feWer tuners are needed to
gather the requisite material (unless the desired programs
Were all on at the same time on different channels). With
only tWo tuners, and the same skim rate and half hour
listening session, the user could pre-store an hour of content
and still have enough programming available for a typical
b. Receive data needed to provide an elec
tronic program guide;
[0030] c. Receive signature or signal ?ngerprint data
used for functions Which are implemented using
program recognition; and/or
[0031] d. Receive metadata for identifying and
“drive” in the car.
describing programs and program segments to facili
tate recording and playback of content desired by the
[0021] Data Storage
[0022] The data storage memory capacity of the RAM
store 105 is a function of hoW often the listener Wants to
change channels. In the above example of equilibrium, (With
four tuners and a 4:1 skim rate) if the listener sWitched
channels every tWo minutes, she Would only need eight
minutes of memory in total. If she Wanted to skim through
half an hour of content before changing channels, she Would
need 30 minutes per channel or tWo hours of memory in
[0023] Finally, the demand for jukebox storage space
might be quite vast to permit the user to tap into large
volumes of radio content from Which to gather songs.
Allocating storage betWeen jukebox and radio listening
[0032] PoWer Supply
[0033] Due the importance of recording scheduled record
ings even When the listener is not present, it is important that
the unit have access to a car’s poWer supply When the engine
is off, as illustrated at 121-124 in FIG. 1. Alternatively, the
unit may incorporate its oWn battery supply (recharged from
the car battery) so that poWer Will be available When it is
needed. The OFF/ON condition of the ignition sWitch, and
other operational data from the automobile’s control system,
may also be used to advantage to control the radio reception
system of the invention; for example, by detecting When car
is not running or in park, the functionality available to the
operator may be altered so that, for example, complicated
Aug. 28, 2003
US 2003/0163823 A1
menus under the Information button are only displayed
When it is safe to do so. When the ignition has been in an
OFF condition for a predetermined length of time (e.g., 36
hours), the unit may be poWered doWn to terminate record
ing and battery drain so that the battery is not signi?cantly
discharged When the car is not in use for prolonged periods.
[0034] User Interface
tuning knob of station-scanning pushbutton(s) may be used
to select other available channels.
Using the Options button, the user can select spe
ci?c channels as “standard” channels to Which one of the
tuners/encoders is assigned and Which thereafter is alWays
recorded. When the unit is ?rst used, the ?rst tWo FM and
?rst tWo AM presets Will be the “standard stations.” The user
can change the mix of FM and AM standard stations by
[0035] The system’s user interface 111 is connected to
control the stations selected by each of the parallel tuners
going into the Options menu. Ared LED (not shoWn in FIG.
101 and the processing performed by the DSP 103 in
(and recording) station.
response to the operation user controls, including a set of
pushbuttons 113, each of Which is visually associated With
a descriptive label (either a permanently affixed on or near
each button, or displayed adjacent to each button on the
LCD display 105, permitting each button to operate as a
1) Will light up to indicate When a preset becomes a standard
[0045] To assign a neW frequency to a preset, the listener
simply tunes in the neW station, and presses and holds the
appropriate preset as she Would With a normal car radio.
“soft-key” Whose function in a given operating mode is
[0046] The Audio Track and Macro-Navigation
shoWn on the LCD display).
[0047] In one embodiment, four presets Will be connected
to four tuner/encoder sets, all Which Would be continuously
In one mode, each pushbutton may be used by the
listener to select programming from a particular station in
and simultaneously recording content When the car’s igni
the same Way the pushbuttons are used on a conventional car
tion sWitch 124 is turned on. In addition, When the ignition
sWitch is off, the unit Will record programming on a sched
uled basis preset by the user. If enough poWer Were avail
able, the recording into the buffer could be continuous, so
that Whenever the user got into the car, there Would be
buffered content available, permitting a predetermined dura
tion of prior programming to be replayed on any one of the
selected channels.
In another mode, the ?ve buttons may be used to
selectively perform the folloWing ?ve functions not avail
able on a conventional radio:
Pause: suspends the reproduction of the live or
recorded programming currently being played. If the
pause button is pressed during live programming,
that programming continues to be recorded for later,
[0048] Each assigned channel may be recorded continu
ously, or When the ignition sWitch is on, by operating the
time-shifted playback. Note that there is no distinc
RAM store 105 as a circular buffer as described in my US.
tion betWeen pausing and turning the radio off, since
Pat. Re. No. 36,801 issued to James D. Logan et al. entitled
the radio continues to record and so is never “off” in
the normal sense.
“Time delayed digital video system using concurrent record
[0039] Save: permits the listener to save the complete
content of a live program currently being repro
[0040] Jukebox: sWitches the unit into “jukebox”
mode in Which the listener can select and playback
previously recorded programming;
Mark: alloWs the user to “bookmark” a spe
ci?c position on a program being recorded (or pre
viously recorded), thereby permitting the listener to
easily return to and resume playback at the marked
[0042] Options: permits the user to obtain informa
tion about available programming, or to perform less
frequently used functions.
[0043] Due to the enhanced navigation capabilities of the
system, feWer settable program selection buttons (“presets”)
selections are required. In addition, there should be a dimin
ished need to surf from channel to channel, particularly
outside the stations that are being recorded. When in radio
mode, then, the skip and scan buttons Would not sWitch to
different stations, but Would resume playback at different
bookmarks Within the recorded content of a given station.
(The unit’s skip and scan buttons Would retain their con
ventional functions When the unit is used as a CD player.) In
radio mode, to change stations, the pushbuttons Would
operate as conventional station selection presets, and a
ing and playback.” In this Way, Whenever the selected station
is broadcasting, the most recently broadcast material Would
be stored, continuously overWriting the oldest programming
from that station. The duration of the most recently broad
cast programming available from each selected station might
be independently selected so that, for example, a station
selected for its frequently repeated neWs and Weather broad
casts might be allocated only one-half hour, Whereas a
station Which broadcasts a favorite genre of music might be
allocated 24 hours.
[0049] Settable selection buttons (“presets”) may be
assigned a frequency the Way they are on most conventional
car radios—by tuning in to a neW station and then holding
doWn the button for a couple of seconds. These ?rst four
presets could be assigned to “standard stations”. Listening to
stations outside of the selected four Would be “roving”, and
could be done under one of four scenarios:
[0050] (1) A?rst approach Would employ four encod
ers and ?ve tuners. There Would not be any recording
or buffering of a non-top-four station. The four
encoders Would stick With the four tuners covering
these presets and the ?fth tuner could rove to other
stations to other stations but not record them. This
approach is useful When the roving is transitory, as
the only use for such buffered recording is to replay
something if the listener missed it. The unit Would
not build up any content ahead of the point currently
listened to, so it Would not be possible to skim
Aug. 28, 2003
US 2003/0163823 A1
[0051] (2) Another, perhaps simpler, model Would
employ the same number of tuners and encoders and
done in one complete action once the program is ?nished, or
done in stages as the shoW is listened to. A multi-minute
reserve one extra encoder/tuner set for roving. This
delay before placing the recorded content in the buffer may
extra reserved encoder Would record and buffer
content from of the “last-roved-to station” When the
be used to alloW the user to jump back While listening to the
user goes back to one of the standard stations.
[0052] (3) Another variant on this Would be to offer
a simple and short buffer for the roving station.
Under this plan, the roving station Wouldn’t store
long term content, but rather provide only “VCR
like” functions (pause, instant replay, etc.) Within a
[0060] For macro-navigation, a long press of the bi-direc
tional skip button Would take the listener to the beginning or
end of the station’s buffer. This approach may be made more
granular (When suf?cient memory space is available) so that
bi-directional skips resumes playback at to the last “major
bookmark”. These major bookmarks Would include:
limited duration buffer.
[0053] (4) The simplest approach is to assume that
only the four standard stations can record. If the
listener roves outside of those there is no recording
and no buffering.
Scheduled programs may be logically placed in one
a. The beginning of the audio track for each
preset (the oldest material);
[0062] b. The end of the audio track (the “Live”
playback position);
c. The beginning of each scheduled recording;
of tWo places as presented to the listener.
[0055] The simplest model Would have scheduled record
ings inserted on top of the respective station’s buffer of
[0064] d. The beginning of a segment of buffered
content (of it at least, say, tWo minutes duration);
recorded content. In other Words, the stored content is
arranged in order of the time it Was recorded and sorted by
e. The Last Listened-To Point (LLTP).
station (and accessed via a selected preset button). The
combined content fro a given tuner (station) is called an
“audio track”. As a result of this combination, no special
her car, gets on the road a minute or tWo later, and then
access means Would be needed to get to scheduled pro
grams—that Would just be there imbedded in that station’s
audio track. If the buffering is going on continuously or after
the recording is made, the content that comprised the sched
uled recording Would not be purged as the circular buffer
rotated but Would remain on the top of the audio track; that
is, programming Which has been scheduled for recording is
protected against premature erasure from that station’s cir
Using the LLTP bookmark, When the user gets into
sWitches channels, the unit does not jump back tWo minutes,
but instead jump back to the beginning to the shoW last
recorded on that channel. This is similar to the Way CDs
operate Where, if the listener skips backWard before getting
too deep into a song, the unit resumes playback at the
beginning of the previous song (track). This LLTP bookmark
is useful When the user is running errands and gets out of her
car frequently. When the user returns to the car, playback
may be resumed.
cular buffer.
This ?rst audio track model also solves the problem
of listening to a station While a scheduled recording is in
progress. With the merged audio track solution, if the
listener schedules NPR to record from 7-9 a.m., and listener
Different audio tones may be issued as cues to
indicate the type of boundary that has been jumped to during
navigation. Alternatively, an audio announcement could be
made, at least initially as a Way to train the user What the
audio cues mean, and then the announcements can be
tunes in at 8:30, the scheduled recording Will merge With,
and actually be the same as, the neWly buffered material. The
scheduled recording Will then be at the top of that station’s
audio track.
cancelled by the user.
[0057] The second model for storing recorded programs
Would have separate storage for recorded shoWs. Separately
user to navigate betWeen songs and talk, and from song to
recorded programs may even be accessed under a separate
segments that are to be saved via the Save button, as
described beloW. Music and voice bookmarks may be gen
“program” button, Which could be used to access and record
programs. To solve the problem of What happens When the
listener gets in the car and a shoW is being recording off the
station being listening to, the unit may store portions of the
program in both the buffer as Well as in the designated
program storage location.
[0058] Erasing scheduled recordings may be done in sev
eral Ways. For regularly scheduled programs, the last install
ment is overWritten When the neW one is recorded. If the
program Was a one-time recording, it is overWritten, With the
space going back into the general buffer pool, as described
later, after it has been surfed or listened to in its entirety. The
content gets put into the oldest part of the buffer (i.e.,
becomes eligible for erasure) to accomplish this overWriting.
[0059] The removal of the overWrite protection and place
ment of a program in a recycle bin of the buffer could be
[0068] Micro-Navigation and Client-Side Markups
Bookmarks are particularly useful to permit the
song. The same bookmarks may also serve to delineate
erated in the folloWing Ways:
[0070] (a) Using the programmed DSP 103 to
execute Fast Fourier Transforms (FFT), the stored
program content may be analyZed and classi?ed into
music or talk, placing a marker betWeen the pre
sumed sWitchovers. TWo “discrimination engines”
may be used to continually evaluate the probability
of a piece of content belonging to either class. An
algorithm dependent on both probabilities may then
used to position bookmarks Which delimit talk and
music. In addition, the algorithms also can frequently
discriminate one song from another. Accordingly,
When tWo songs are placed back to back, the algo
rithms construct an estimate of Where the transition
is and creates a dividing bookmark.
Aug. 28, 2003
US 2003/0163823 A1
[0071] (b) “Song ?ngerprints” (song ID technology)
[0080] Adifferent approach from Word recognition (across
can be used to identify songs. The beginning and end
of a song may be determined from data provided
different speakers) that may be employed is the automatic
recognition of repetitive patterns on talk and neWs shoWs
With the ?ngerprint data Which specify the time
that are used over and over in the same program or on
durations Which separate the start and ?nish of a
multiple days. These sounds may Well be part of the shoW
song (or other identi?able program segment) from
the time position of the detected ?ngerprint. A
format. Repeating patterns can be detected and saved. For
delayed further offset could be used to account for
the usual DJ talkover at the beginning of a song so
that the bookmark might be set into the song some
What so that, if talk is detected after the expected
example, a neWs station may repeatedly announce traf?c
reports With a phrase such as s “Traf?c on the Threes.” Such
patterns repeating patterns can be detected used to determine
bookmark positions that are useful for navigation and seg
ment saving. If an audio artifact Was typically at the begin
ning of an ad block (“We’ll be right back after this”), a
start of the song, the delayed offset position may be
bookmark could be placed tWo minutes or more after the
[0072] The ideal marking technique Would combine both
of these methods, and thus offer several advantages:
[0081] Different voices (different speakers) may be
[0073] While the classi?cation method can mark off the
DJ talkover nicely, it does produce errors; i.e. false markers
and missing markers. False marker errors associated With the
classi?cation method could be detected With the Song ID
technology and the false markers deleted.
[0074] The Song ID approach could also highlight the
presence of a missing mark. Such detection could prompt the
algorithm to re-analyZe the content in the approximate
location of the missing marker. The re-analysis Would be
detected using speaker identi?cation analysis to place a
bookmark When a neW voice appears, When a voice reap
pears after a certain time period. If the host’s voice is knoWn,
a bookmark may be placed each time a voice that Was not the
host’s appears. If music is detected in the background on a
talk shoW station, the segment may be assumed to be an
advertisement. To ensure that these, and other auto-gener
ated bookmarks, do not appear too frequently, only using the
?rst instance of closely spaced similar detected event Would
be used as a bookmark.
done With neW parameters that re?ected that fact that there
Was a marker in the vicinity. If no split Were detected, the
bookmarking mechanism Would perform the split, for
example by averaging out the tWo split points deduced from
[0083] Metadata, including predetermined time positions,
each song.
[0075] The bookmarking mechanism eliminates the need
to run all the music content through the splitting process,
since only those areas thought to be close to a split need be
Time based bookmarks may be used to identify
normal program breaks, such as on the hour and half hour.
?ngerprints and other audio patterns for music, individual
speakers, repetitive events, and other detectable bookmark
events, may be doWnloaded to the unit in a variety of Ways
as previously noted. In addition, as discussed above, book
mark events may be detected by signal analysis. Events
designated by both metadata and by signal analysis may be
used in combination to determine bookmark locations.
As there is a lot of music in advertising blocks, the
When a shoW is ?rst scheduled, the system could inquire
music bookmarks Will only be set When a solid block of
music equal to or greater in length than the minimum song
bookmarks. When using speaker recognition techniques,
length, at least tWo minutes, Which is much longer than the
music content found in advertising.
Other forms of micro-navigation the provision of a
30-second (or other predetermined duration) skip button,
and the use of analog fast forWarding (preferably While
playing a listen-able time scaled version of the audio, and
skips to micro-bookmarks.
Whether the user Wanted the shoW to contain any time-based
speech samples of the knoWn speakers on each shoW may be
provided in order to construct bookmarks When speakers
change. The system may also deduce Which bookmarks and
types of marks are of value over time by observing Which
type of marks any individual user makes use of.
The system can further analyZe usage logs from an
Micro-bookmarks for talk shoWs may be produced
aggregation of users to determine Whether people are not
making use of a particular bookmark. When a signi?cant
number of users don’t skip right after having landed on a
by using speech recognition to identify speci?c Words such
playback point designated by a given bookmark, it may be
as “neWs”, “sports”, “traf?c”, etc. This approach may lead to
too many false marks, hoWever, as these Words may be
overused in the content and Would have to be recogniZed
across a large number of occurrences. Selecting only the ?rst
instance in a cluster Would help minimiZe this problem.
assumed the bookmark provides a useful demarcation point;
otherWise, the bookmark may be deleted from metadata
provided to later listeners.
Other Words or phrases to look for Would be
segueing phrases (“Thanks for calling”, “Welcome to the
shoW”, “When We return”, etc.). Other sounds Which could
be detected to create bookmarks include laughter, applause,
croWd noise at a baseball game, silence, deep breaths and
[0085] Bayesian probability techniques can be used to
re?ne the use of auto-generated bookmarks. In particular, the
time duration betWeen breaks may be an important piece of
data indicating the probability of another break coming up.
Programs may employ a frequency distribution indicating
hoW often separate segments identi?ed by automatic book
marking occur. As time progresses With no bookmark indi
person’s speech may also be determined to provide infor
cating a change in topic, the more likely a change is to
happen soon, and the parameters in the search algorithms
mation used to separate program segments. The bookmark
might be best placed a feW seconds before these sound
[0086] A user may adaptively train the system to look for
the right repeating sounds to use as bookmarks. Training is
other sound effects. The excitement level or speed of a
may be adjusted accordingly.
Aug. 28, 2003
US 2003/0163823 A1
performed by creating bookmarks at locations Where a
recurring bookmark Would be desired and erasing those of
limited use that the system had automatically created. The
creation and deletion of bookmarks is more fully discussed
[0087] Note that many or all of these bookmarks could be
embedded in a digital or even analog radio signal With the
cooperation of the broadcaster. Emerging satellite and digi
tal terrestrial broadcasting systems Which can be used to
particular advantage With the program storage and playback
tising be left in, or that certain songs only are alloWed to be
played a certain number of times, or that less popular songs
are played along With more popular ones.
[0097] Various subscriptions fees could be associated With
various uses.
[0098] If acceptable business rules Were agreed upon, the
radio stations may supply the metadata, or alternatively the
record companies may embed the data in the recordings
capabilities provided the unit may imbed bookmark infor
mation and other metadata in the signals transmitted to the
stations, in conjunction With the record companies, broad
casting songs (preferably late at night) to permit recording
[0088] Audio Tracks
[0089] The system may advantageously utiliZe “user-gen
erated audio tracks” Which users create by editing recorded
audio tracks. In this system, the system Would record long
strips of unmodi?ed audio, or audio With detected talk
removed to form a continuous music recording. The listener
Would then listen to recording, using the available analog
navigation tools. When the user Wanted to create a book
mark, an input signal from the user Would be accepted. After
one or tWo listening sessions, the most desirable segments
Would by identi?ed and delimited by these user actions. The
bene?t over the automated markup mechanisms described
Another business model Would have the radio
and collection at the client side by the recording device. This
model could be a means for “distributing” CDs to paying
[0100] Where to Start Listening
What happens When a user leaves a station and
comes back to it? Either because they paused the system,
changed channels, or left the car?
[0102] One option is that listener alWays starts at the live
radio playback point, as they Would With a normal radio.
While this similarity has some bene?t, starting With live
programming Would make it more tedious to go back to the
above is that the unWanted songs are never designated in the
point at Which the user paused.
?rst place. User-generated audio tracks require greater effort
[0103] A second option is to start up at the last (previous
in time) major bookmark. If there Were no recordings that
occurred in the interim, that Would take the listener to the
LLTP. For short breaks, this option does exactly What Would
be expected if the listener paused the station. If the listener
by the user, but are most useful to Willing participants Who
enjoy the creative effort required.
[0090] Another Way to get to roughly the same result
Would be to let the user surf through, and listen to, the raW
audio track. The system Would then start to place bookmarks
in those places Where the user actually listens at normal
speed to a segment over 30 seconds long. In this Way, the
system semi-automatically determines Where the good songs
and other desirable program segments are recorded.
In another model, the bookmarks Would be
deduced by aggregating and averaging multiple bookmarks
from multiple listeners thus creating community bookmarks
is out of the car for a While and a scheduled recording
occurs, then the unit Will start up at the beginning of that
scheduled recording. If the unit has not been playing a given
station for a predetermined time (say, 20 minutes), and there
is no scheduled recording in the audio track, playback Will
begin at the beginning of the audio track (Which at that point
Would be operating as a buffer holding a predetermined
duration of the most recently received programming. To get
to live radio, the user Would traverse forWard just once using
to be used for the same purposes above.
the major bookmark button.
[0092] Under either scheme mentioned above, the system
could start to delete content not of interest or not being
listened to in order to conserve storage space. It Would be a
form of “use it or lose it”.
an experienced user may Want is to start sur?ng through
In place of a jukebox, the system could save
Rather than starting at the live or LLTP positions,
recently stored material, not yesterday’s content found at the
LLTP, Which Will be too old. Going to live, on the other
hand, may often be too sloW paced for experienced listeners.
bookmarked and edited audio tracks as discrete units. Users
could access and play speci?c audio tracks (With the neW
bookmarks added and unWanted material deleted over time)
constantly recording 24 hours a day for some or all the
recording stations. By using devices Which consume little
on demand. Tools may be provided to permit the user to
poWer, sustained operation using the car’s battery is practi
reorder the segments (songs) in the audio track (but not to
exchange segments With other tracks to avoid confusion).
cal, particularly When a time limit is included to poWer doWn
[0094] The CD player/recorder seen at 109 in FIG. 1 may
be used to alloW the listener to burn CDs Which preserve the
edited audio tracks.
[0095] A Cooperative Business Model
Content may be placed in a large buffer that is
the device after an extensive period during Which the battery
is not recharged. As a result, the buffer is alWays full When
the user gets in the car. This buffering should be performed
in a Way that does not over-Write any scheduled recordings
stored in the audio track or content surrounding the LLTP
from the day before.
[0096] The device may be adapted to support negotiated
business rules that Would control the usage and construction
probably be the LLTP as users Will seldom schedule music
programs or care about the time-of-day When listening to
music stations. As user’s may skip over much of the DJ talk,
of the audio tracks, the jukebox, and resultant CDs. These
rules might mandate a certain amount of the original adver
For music stations the last major bookmark Would
Aug. 28, 2003
US 2003/0163823 A1
and just go from song to song much of the time, any
the above-noted applications. Under this plan, people could
out-of-date comments made about traf?c or Weather, etc.
mark up shoWs using their oWn listening softWare and
upload the resulting metadata to a shared server for doWn
Will be skipped. Thus by following the rule of starting at the
last major bookmark, the starting point for a music station
may be the left-off point from the previous day.
[0107] Regarding another question relating to pausing,
attention should be given to the issue of What to do When
someone leaves the car. (Thus solving that age old conun
drum: Do I sit in the car and listen to the rest of this song or
talk shoW segment, or leave the car and forget about it?) The
simplest approach may be to record additional material from
the station being listened to if the listener is 15 ?fteen
load to others.
[0112] The audience itself could generate community
generated bookmarks if there Were a back channel (a client
to-shared-server link) to collect them. With such a back
channel, the shared server metadata may identify “Hot
Spots”, a guide to the most popular segments. In addition, by
monitoring listeners’ use of navigation controls and time
scaling, and analyZing such usage patterns, useful book
marks may be automatically generated for the bene?t of
minutes or less aWay from live listening. So if the user Were
other users Who listen to the same programming at a later
listening live When the car goes off, another 15 minutes
time. As many people Will presumably be using time scaling
Would be recorded. If the user Were listening 5 minutes aWay
as a Way to skim until they come to something interesting,
from live, the system Would record another 10 minutes of
content. In all cases, the listener Would alWays have avail
able 15 minutes of content from the LLTP of the last station
the analysis performed at the shared server can identify
those locations Where people tend to be sloWing doWn to
listened to.
bookmarking metadata. These deduced bookmarks Would
[0108] Whether to startup again in Live or the LLTP is a
decision that could be left up to user Who could make such
a selection in the preference menu. Alternatively, the On/ Off
button Which pauses the radio, could be designed such that
pushing it in and out turned the unit off and paused it (and
so turning it back on Would bring the playback point to the
LLTP), While turning the button counterclockWise Would
shut off the radio, in Which case the listener Would Wake up
in live radio. In this Way the user could make the decision
on the ?y.
[0109] Ways to Listen, DoWnloaded Markups
listen carefully and designate those preferred segments With
then be doWnloaded to later listeners.
As users can create their oWn bookmarks, these too
could be collected for shared use using the back channel.
When consistent usage patterns are observed, these user
generated marks could be passed along to later listeners.
Other Ways to Listen that are akin to the “con
densed shoWs”, “previeWs” of shoWs (audio trailers), and
“highlights” of talk shoWs described in the above-noted
previously ?led patent applications.
[0115] Voice recognition may be used to construct the teXt
of spoken Word content. This teXt may then be analyZed for
the same Way that it Was used to control television program
meaning With AI tools and a summary extracted. This
summary may be displayed on the LCD 115 or spoken via
playback as described in the previously identi?ed patent
a teXt to the audio output.
[0110] Metadata may be employed to control playback in
applications including US. application Ser. Nos. 10/060,001
and 10/165,587. Using these mechanisms, timely metadata
may used to identify and categoriZe segments of recorded
talk or neWs (Which could optionally be run through a
preference ?lter). The user may playback identi?ed seg
ments continuously, or use the metadata to surf. The result
Would be a personaliZed neWs shoW. The metadata may be
supplied in one of several Ways: It may be pushed to the
client playback unit either through a point-to-point Wireless
[0116] Audio content may be previeWed by playing back
highlight segments in a scan mode Where portions of each
highlight segment (based on time or content-based book
marks) are played for a feW seconds. An audio indeX could
be constructed by using metadata to capture a key sentence
from each segment to be used in the audio indeX. The system
could read these in a predetermined order and the listener
could randomly access each in turn through a voice com
mand or button action.
link, embedded in a broadcast stream, or transferred to the
system in batch mode such as via a CD or ?le transfer from
of a Palm Pilot. In either case, the doWnload vehicle Would
[0117] There are also novel Ways that music may be
played back. For instance the musical equivalent of a
not knoW exactly What had been recorded and What metadata
television “favorites” playback may be constructed by speci
Was needed. Available metadata (or a subset that a speci?c
fying the musical group (or speaker for a talk shoW) the user
is interested in hearing and the system could assemble those
songs or segments for automatic playback. This may also
user might need) Would be supplied and the client softWare
Would select the needed metadata and match it up With the
content. For a push-Wireless system, a broadcast carousel
system in Which the metadata streams by and the client
softWare select the needed elements, may be used. The
metadata may be embedded into the FM sideband of another
channel. With a multituner system, a loW-priority tuner
Would be used to go pull data of this stream as needed to
bring in the metadata. A further method places the metadata
in an IP-over audio stream.
[0111] Note, by employing a doWnload capability, the
form a useful Way to rearrange the listener’s saved selections
(jukeboX)—by artist.
[0118] Another variation in music playback Would involve
modulating Where the marker for the start of each song is
placed. A user preference can be set to tell the system
Whether the user Would prefer to have the marker after the
DJ talkover had stopped, or to have the marker placed Where
the music is ?rst heard in the background (even though the
DJ is still talking), or at an even earlier point providing for
metadata to Which identi?es and describes neWs and music
more of an intro. (This Was discussed above Where it Was
segments (applications called “NeWsCatcher” and “Song
Catcher” respectively) can be supplied by the listening
public under the “community markup” methods described in
suggested that the user might Want to land Within the song
so as not have to go through the talkover.) If the marker Were
placed after the DJ talk, and the user decides he or she Wants
Aug. 28, 2003
US 2003/0163823 A1
to hear the Whole song even With the DJ talkover, they could
hit the back skip button and it Will go to the location Where
the mark might have been under the other decision rule at the
very beginning of the song.
“Options” button (one of the multi-function, user-actuated
buttons indicated at 113 in FIG. 1). These functions may
include but are not limited to:
[0126] (a) Scheduling recordings and changing the
[0119] A “scan” button may be used When navigating
time-shifted content. In operation, the scan button When
[0127] (b) Rearranging playlists;
operated moves the playback point to each bookmark in
[0128] (c) Preventing the overWriting of stored con
[0129] (d) Perhaps labeling songs if there is no Song
sequence and lets the listener set the amount of content
played after the bookmark, the length of Which might vary
depending on Whether the content Was talk or music, and
What type of bookmark Was encountered. If the time
betWeen bookmarks Was too long (for instance, With a
talk-shoW for Which no markup metadata Was available),
ID available; and
[0130] (e) Setting various preferences. While differ
then a skipping pattern using preset time jumps, or jumps to
automatically detected audio events, could be employed to
ent preferences could be set for different occupants
alloW the user to scan through the content of a recorded
to different content. As such, the same effect can be
of the car, chances are the different occupants listen
audio track.
achieved by offering preferences for different shoWs,
[0120] Another variation on the scan function could take
the listener to a point in time in the segment that is after the
bookmark. This Would give listeners a quicker sense of What
stations, etc.
the segment Was about than starting at the beginning of the
segment. (This in essence is the Way a conventional radio
scan button operates, by taking the listener relatively deep
into the content by virtue of the fact that it is non-intelli
gently jumping into a broadcast stream in process in the
channel (station) jumped to.
[0121] Songs are knoWn to have a “sWeet spot” in
them—a spot that resonates With, and is quite familiar to
listeners. It is the segment that often is represented on
sampling Websites Where listeners can hear a 30 second
[0131] While accessing these functions through a menu
system could be someWhat tedious, the actions done not
need to be done often.
The menus could be presented in any of the fol
loWing Ways:
[0133] The menu options may be read aloud via the audio
output to communicate a list of choices. The options may be
presented either in a timed sequence, requiring the listener
to respond at the right moment, or by invoking the reading
of each choice With a button action, providing the user With
audible feedback on choices made Without the need to Watch
snippet of each song. Quite often this segment Will be the
the LCD display.
refrain and include the name of the song as part of the lyrics.
[0134] Alternatively, the menu may scroll through a list of
choices, Which Would be presented on the LCD, using the
bi-directional seek button.
SWeet spots may be used in another form of “scan-listening”
in Which the playback jumps to the sWeet spot in each song.
If the user likes the song, he or she can hit the back skip
button to go to the beginning. This concept could be used
both for buffered music content and With songs saved into
the user’s jukebox. The sWeet spots may be identi?ed by
doWnloaded timing metadata or isolated using the DSP to
perform song identi?cation.
[0122] When a song is played back, and the Song ID
system is employed, the song name and artist Will be
displayed on the LCD 115. To supplement that, the system
could use a text-to-speech system to say the song and artist
names so the user Wouldn’t have to take his or her eyes off
the road. Alternatively, audio clips of someone pronouncing
The user could select an offered menu item by
either hitting a button, or using voice input. A simple voice
system may be used that accepts simple replies such as “yes”
or “no” from a range of users. Alternatively, the system
could be trained to understand the operator’s voice, as some
cell phone systems do today. If the voice command recog
nition system simply recogniZed letters and numbers, voice
input could be used to input data by spelling it aloud. Amore
advanced system could dispense With either the presentation
of the menu choices or spelling, and alloW the listener to
speak the names and times to be provided as input, as is done
in the voice recognition systems used by some call center
the song and artist names could be doWnloaded for better
and message handling telephone interfaces. A cell phone or
audio clarity.
other bidirectional connection may be used to connect the
client unit to a robust voice recognition and command
system operated as a shared telephone, a Web service or the
Given the up to 10:1 ration of storage needed for
spoken Word compared to music content, storing these
names Would be an insigni?cant amount of overhead if it
Was only being stored for recorded songs. Storing the
identifying sound clips for the Whole database Would require
extensive and expensive storage capacity. To conserve
space, the metadata server may doWnload the audio name
clips and other descriptive metadata only after recorded
host metadata service, Which Would return command signals
and/or metadata for controlling the client unit.
[0136] Another non-menu-based method to input data
takes the form of a “scrolling alphanumeric keypad”. Under
this system, the user could use the scan button to scroll
through a list of numbers and letters on the LCD screen and
songs Were identi?ed at the player and their identity sent via
select each character one by one, using the buttons 113 as
the back channel to the metadata server.
“smartkeys,” in the same Way that users input their name or
initial into the “Highest Score” screen of a video arcade
game. For numbers and specifying am. or p.m., this scroll
[0124] Menu Functions
[0125] Enhanced features could be accessed through
ing alphanumeric keypad Would be relatively fast and pre
Menu functions that Would be accessed through the
Aug. 28, 2003
US 2003/0163823 A1
[0137] Another potential interface Would piggyback on
the cell phone interface. If the cell phone keypad Was
interfaced to the system, for example using a Bluetooth®
link (see, the needed alphanu
meric data could be typed into the cell phone using its
keypad and display screen, With Which the user Would
already be familiar. Other portable devices Whose in-car
interfaces the system could piggyback off of include Palm
Pilots, and notebook and tablet computers, Which could
communicate With the system via a USB port, a Bluetooth®
that station. The recording could be associated With the
originating station (using a LED to indicate the presence of
the recording on that normally-not-recording station). Leav
ing it there Would not require any additional tuners or
[0151] Alternatively, the recording could be placed in the
audio track of any one of the recording stations as if that
station had recorded it.
or other radio link, or an infrared connection.
[0152] While the model of recording a program off a
non-recording station is understandable for content recorded
[0138] If the system had the bene?t of exchanging data
during non-driving time, it could be confusing When driving
With a host computer via an MP-3 CD, ?ash module, or a
as one of the normally recording stations Would not be
Wireless netWork connection, menu selection could be done
on the host computer and the results transmitted live to the
recording (as that tuner/encoder Would be temporarily
devoted to recording the non-standard station). The user Will
unit, or doWnloaded in advance as a playlist to the car unit.
appreciate the consistency of alWays having content from
Making choices using a PC Web interface Would be provide
the same four standard stations being buffered. Cutting out
substantial ease-of-use, safety and ?exibility advantages.
one encoder to go pick up a shoW from another station Would
Alternatively, a robust interface, such as a touch screen
break that paradigm, forcing the listener to choose Which
LCD, provided by the unit itself Would facilitate the menu
presentation and selection being made in the car.
encoder to use. It Would be especially confusing if We used
[0139] Depending on safety concerns, certain functional
ity should only to be used When the car is not being driven
[0153] Assigning a recording to go into the audio track of
the originating station or that of another station: This
approach Would alloW someone to put all his sports shoWs
in one audio track and political talk shoWs in another,
and Would be inhibited When the car is in motion (or not in
[0140] Scheduling Recordings
[0141] Scheduling a recording may be the most important
function requiring a menu interface. A speci?c example is
outlined beloW to serve as an example of hoW the menu
system preferably operates.
To schedule a recording, the user Would:
[0143] a. Select the Schedule Recording option from
Within the menu presented under the Options button
the roving tuner/encoder approach.
creating in essence, a “virtual channel”.
[0154] If the separate Program button is used, or if the
menu routine had enough options, programs could also be
stored in distinct bucket, one or more for each preset buttons.
In other Words, the preset buttons could access bins Within
Which to store and retrieve scheduled recordings. Again,
these could serve as virtual channels.
[0155] A recorded program Would be given some protec
tion from being overWritten When it is put into a station’s
or under the Programs button if that Were present.
audio track. This Would alloW the listener to record her
[0144] b. At the audio or visual prompt, specify
Which standard station Will be recorded by hitting
that preset button.
favorite gardening shoW from Saturday, and listen to it
during the Week Without it being overWritten With day-to
c. Select a beginning time and end time from
the menu of choices.
[0146] d. The system Would announce (audibly or
visually) hoW much time Was left for other purposes
and the listener Would have a chance to undo the
scheduled recording.
day material. When invoking this feature, the system Would
inform the listener to What extent the capacity of the buffers
had been reduced for the standard stations. One option
Would be to alloW a recording to only be overWritten by the
recording of the next episode only, as mentioned above.
[0156] To explore programming content from times of day
that aren’t normally experienced (but during Which desirable
Given this fact, and the fact that scheduling isn’t done
frequently, more complex features may be provided for the
“poWer user” invoking this feature.
programming may be recorded and time-shifted to the times
When car is being used, and given that there is no currently
available EPG (electronic program guide) for radio, a
“recording scan” function may be implemented. This
mechanism Would record the ?rst 10 minutes at the top of
the hour for the Whole day for a certain station. From this
sampling, the user may elect to play back certain recordings,
and the identity of these preferred programs thus be adap
tively learned for later, more complete recording the fol
loWing day or days.
e. Choose a “frequency option” from: this day
only, everyday, all Weekdays, individual recurring
days of the Week, or Weekends.
[0148] Scheduling recordings should be a car-off activity.
Other scheduled recording features that could be
made available, at the risk of further complication, are
described next.
Creating a scheduled recording from a non-stan
dard station: Under this scenario, a station to Which the user
did not Want to devote a tuner/encoder set might have
individual programs Worth capturing. The system may
sWitch a tuner/encoder to that station at that time and record
With the bene?t of being able to doWnload EPG
metadata in a point-to-point manner as previously described,
the users could use a radio EPG on their PC to make future
recording selections, With the selections being transferred to
the unit to control recording. The same PC interface could
also be used to operate other functions of the radio, as noted
earlier. Alternatively, an LCD touch screen display on the
Aug. 28, 2003
US 2003/0163823 A1
in-car unit may display menus containing downloaded EPG
metadata, and the recording selection then made from that
easier-to-use interface.
The Save Button and the Jukebox
[0170] Novice users Would hit the Jukebox button (and not
hit a preset button after that) and the linear list stored under
preset #1 Would start playing from neWest to oldest. The
second-mentioned novice system Would play the linear list
associated With the station the listener Was last on before
[0159] When the Save button is invoked, the content
currently playing, either a song or other content, Would be
hitting the jukebox button.
saved in one of several Ways:
system Will start playing from the playlist represented by
that preset. (The “unassigned linear list” is represented by
[0160] If live content is playing (and no bookmarks mark
ing the content have yet been created), the system Will save
a predetermined amount (eg 10 minutes) of content sur
rounding the point When the Save button Was pressed (e.g.,
the last ?ve minutes and the next ?ve minutes to be played).
Bookmarks for that content Would be created, and the saved
If a preset is hit after hitting the Jukebox button, the
preset #1.) If another preset is hit at any time after entering
jukebox mode, that playlist Will start playing.
[0172] Whenever a playlist is accessed, it Will start play
ing from the LLTP as if it Were a CD that had gotten shut off.
As mentioned above, playlists Will alWays play from neWest
segment could be trimmed later as described beloW.
to oldest.
[0161] If the Save button is hit While playing a recorded
segment that has been bookmarked, the unit Will save the
[0173] The Jukebox Will also be able to employ the
“random” and “shuffle” buttons if they are present in the
system. These commands Would randomly play or tempo
rarily shuffle the songs in a given playlist.
segment de?ned by the previous and next bookmarks.
[0162] In a model Where there Were be no bookmarks
delineating songs, the user may Would hit the Save button at
the beginning and end of the section that Was to be saved.
Audible or visual feedback may be used to indicate to the
user that the ?rst Save press indicated the start of a recording
and the next Save action marked off the end.
Songs may be saved and assigned to the “Jukebox”
[0174] A limitation may be imposed on hoW long a
segment could be that Was not manually trimmed (see about
trimming beloW). That Would prevent overly long segments
from being dropped in by mistake (for instance, four back
to-back songs that Weren’t split apart With bookmarks.
A method to doWnload data, such as a CD-RW
in one of three Ways:
player, ?ash module, other removable media, the USB port
[0164] The simplest implementation for saving songs is to
player, a connection to a PC, etc., opens up the opportunity
to have a jukebox that could include other music beside What
had been collected off the airWaves. For instance, the user
could put a CD into the player and hit the “save” button if
she Wanted to transfer a given song and store it in the
Jukebox. Songs from MP3 players could be transferred
digitally. The result Would be a jukebox With songs off the
or a Wireless connection to a portable CD or MP3 music
simply hit the save button and the song (or the segment)
being played Would go into a single linear list of songs.
[0165] A step up in functionality Would have each record
ing station have its oWn playlist. In this case, When the user
hit the Save button, the song Would go into a linear list
(organiZed by date saved) associated With that station.
For more advanced users, the mechanism Would
alloW a user hit the save button and then immediately hit a
preset button. This Would assign the song to a speci?c
playlist of the user’s choosing. In the simplest novice case
above, Where no preset Was hit after hitting save, the song
Would go into a default linear list, Which Will be assigned to
preset #1. In the second novice approach suggested above,
the saved song Would go into the linear list assigned to that
respective station.
[0167] In all cases, the “original copy” of the saved clips
radio, CDs, or MP3 players.
[0176] Alternatively, selected functions implemented by
the receiving system described in this speci?cation could
take the form of features in other music systems that might
be in a car, for instance an MP-3 jukebox With a hard drive.
In this case, the reception and storage of broadcast program
ming Would not be providing the “dominant” jukebox and a
means Would be necessary to place songs captured off the
airWaves into that jukebox.
[0177] Amethod to upload content to another computer is
also envisioned using a CD-R, Wireless link, etc. This action
or songs continues to reside in the buffer or recorded
Would be done With a button sequence or voice command
program. Until the original content is purged, there Will be
tWo virtual copies of the saved material in the system (the
While the desired content is playing. If the reception system
displayed a list of songs developed using its song recogni
second copy being implemented by memory pointers Which
identify a corresponding segment in the buffer).
tion system, selected songs could be uploaded or Written to
[0168] To access the jukebox, the user Would toggle the
AM/FM/Jukebox button.
[0169] The system could offer any number of playlists
stored under the preset buttons by having this button toggle
through AM, FM, Jukebox 1, Jukebox 2, Jukebox 3, etc. The
user could set any number of playlists through the prefer
ence settings. As an example, With four hours of storage in
available memory, the Jukebox might store tWo hours of
music (about 30 songs); so one set of presets Would be
a CD-R or ?ash memory card.
Content could be Written to a car-based CD-R or
CD-RW. Here it might be advantageous to only permit this
Writing function if the car is off or stopped to avoid skips on
the laser-based media. The commands could be given at any
time and the Writing could take place later.
[0179] If storage space is not available for the complete
?ngerprint database needed for Song ID then the database
could be broken into tWo parts:
[0180] (1) Fingerprints for the most common songs
heard on the radio during a given period of time, and
Aug. 28, 2003
US 2003/0163823 A1
songs that the user had expressed an interest in either
by saving them or including them on a Huntlist or
features may be incorporated into the jukebox, adding the
Never Again list, Would be stored on the client.
These ?ngerprints could take up to 1K of memory
bookmarks and creates neW ones Which more accurately
ability to “trim” a song. To do this, the user deletes old
delimit desired content.
each so their storage on the client may be limited.
[0181] (2) The rest of the much larger database could
be made available in one of several Ways:
(a) The ?rst Would employ a CD or other
means to supply metadata. This database could be
used to do Song ID in batch mode. A database of
700,000 songs Would be more than suf?cient for
radio broadcasts and could be stored on a single
CD. Each time the CD Was entered into the
system, ?ngerprints for neW songs and other rel
evant songs needed for the database stored on the
client Would be transferred to the unit.
[0183] (b) In another architecture, a data carousel
could be available via broadcast, alloWing a larger
database to be continually streamed to the radio
and used for identi?cation.
[0184] (c) Another architecture (discussed in the
above-noted patent applications) Would involve
sending snippets of recorded songs (via a Wireless
connection) to a server for identi?cation. Song ID
softWare, Which typically scans a complete user
supplied recorded song by a vendor-supplied ?n
gerprint looking for a match, may be modi?ed to
transmit user-supplied snippets to the shared
server, Which compares the snippet With a vendor
supplied complete songs looking for a match. The
supplied snippet is preferably could be taken from
a standard time into a song’s play in order to speed
up the matching process. The latter method Would
also require that talk/music discrimination soft
Ware be used to identify What content material is
[0185] This snippet approach could prove to be useful
When trying to identify hard to identify songs and When it’s
dif?cult to get the entire database to the client. In fact, if
there is a viable point-to-point Wireless connection avail
able, this may prove to be the best Way to build up the
[0189] To create a bookmark simply hit the Bookmark
button. The bookmark Will be inserted at that point in the
audio content then playing. To delete a bookmark, simply
skip to that bookmark and press and hold the bookmark
button Within a second or tWo of skipping to the bookmark.
Note that trimming could be done to material
before or after saving it. That is, the user can use this
bookmark process to setup material to be subsequently
saved. This Would be the means to clip ads, or talk content,
or to cleanup songs (before saving them) that had errant
There could be cases Where a song or clip Was
saved to the jukebox but still needed cleanup. This cleanup
Would be easy using the bookmark tool if the listener Wanted
to shorten the clip. By creating a neW bookmark at the
beginning of the clip, the beginning of the song or clip Would
be moved in to that neW point. By creating a neW bookmark
at the end of the clip, the song Would noW end at that spot.
In some cases, hoWever, the saved song may be too
short. In anticipation of that problem it is proposed that some
extra content be saved outside the bookmarked start and stop
points of a song just in case the user Wanted to move the
bookmarks back.
[0193] To move bookmarks outside the demarcated song,
the user has to have the ability to navigate past a bookmark.
So if a song as bookmarked in the jukebox that “started too
late”, the user may move the bookmark back in time. While
this could be done in several Ways, perhaps the best Would
be to go to the beginning of a song (by hitting the skip button
taking the playback point to the beginning of the next song
in the playlist) and from there pressing the analog reWind
button. Normally this Would not do anything at the begin
ning of a song but in this case, it Would move the playback
point before the extra material stored in front of the start of
the song.
of unknoWn songs being sent to the server, and the ID
[0194] By going backWards, the system knoWs the user is
trying to re-specify the beginning of the song and Will “add
information along With the ?ngerprint for future use being
returned to the client. This approach may require less
material” to the beginning or end of the song. The user can
noW reposition the bookmark Within this neW material in the
bandWidth than the alternative, Which is to doWnload a
standard Way. For this approach to Work With already saved
content, it Would require that the extra material be available
before the beginning and after the end of each saved
database of ?ngerprints needed on the client, With snippets
potentially larger number of ?ngerprints to keep the data
base full in anticipation of What might be recorded.
(d) If space for a ?ngerprint database Were not
a constraint, then the radio Would be sold With a
complete database of all ?ngerprints for all the songs
that have appeared on radio in the last 40 years or so.
segment. This extra material could be saved for a certain
time period and expunged after the song Was trimmed or
after the material had been listened to a feW times With no
attempt made to trim the material.
The supplied data base then only need be periodi
cally updated as neW songs come out. Again, any
number of one-Way connectivity schemes can get
Would be to snip out talk or neWs segments the user Wished
to save. With appropriate connectivity, users could also
this update metadata to the unit, including broad
casting it on a PM subcarrier, cell phone connection,
bookmark content and send these bookmarks (or the actual
Another important use for the bookmark button
content) to other users Who had recorded the same content.
or a CD-RW disk, etc. as noted earlier.
[0187] The Bookmark Button and Song Editing
[0188] If the product is to have a stronger music focus,
given a device-adroit male target audience, more aggressive
To delete a song or saved segment, the user Would
simply start to play the saved segment and hold the save
button doWn for a feW seconds. When pressed and held, the
Save button becomes an Unsave button.
Aug. 28, 2003
US 2003/0163823 A1
[0197] User Interface
tion that might otherWise have to be displayed on the
[0198] The functions performed by reception system
dashboard. In addition to information on request, the car
OEM could use our audio output system to distribute
should be provided by an easy-to-understand, easy-to-use
user interface that does not rely on a user’s manual or on
visual information (so that the driver can access commonly
use features Without taking his or her eyes of the road). To
achieve this result, it is desirable to extend the functionality
of familiar radio/CD buttons as much as possible.
[0199] Audio prompts may be used, but could be turned
off in the Options menu or could go aWay after being used
a certain number of times.
The unit preferably includes, at the minimum, a
small LCD display seen at 115 in FIG. 1. This display Would
shoW station, artist and song name When available, recording
status, and importantly, When time shifted broadcasts are
information When the driver didn’t necessarily request it (a
“push” application). For instance, the system could
announce: “A 15,000 mile service is needed,” or “The right
rear door is not closed.” In addition, the memory and
communications capabilities provided by the unit may be
employed to store GPS navigation maps, to store diagnostic
information from the vehicle’s computer, to store and dis
play route and trip planning information from a connected
PC, to store telephone and address information used by a
cellular phone or PDA, etc., to collect and display traffic and
Weather information from local radio sources, to use its
Wireless connectivity to access Web services of use to
travelers, etc.
being played, the relative buffer location being read relative
[0214] Memory Allocation
to the “live” location.
[0215] The system Will also automatically adjust some of
[0201] Atime bar Would express relative listening position
the uses of memory over time, although nothing Will ever be
both With numerals as Well as With a dynamic, variable
length bar in Which the Live playback point appears as the
deleted from the jukebox automatically. While a standard
buffer siZe for all four recording stations Will initially be set,
right-most point and the listening point positioned to the left.
the length for each station could be automatically adjusted
The bar could represent either the relative position of the
listing point in relation to the Whole bar Which represents the
total recording in the buffer, or it could represent absolute
time (one hour, tWo hours, etc.). For playing back a Sched
uled Recording, the time bar Would not represent distance
from Live. Instead, the full bar Would represent the length of
the entire shoW and the moving marker Would shoW Where
over time. The system may automatically adjust the buffer
siZe by adaptively recogniZing usage patterns. For a station
broadcasting neWs, for instance, Which repeats most of its
in the program the listener Was at that time.
[0202] As noted earlier, digital devices that have limited
and restricted interfaces Which employ useful interface
techniques that could be incorporated into the reception
minutes. The unit may “learn” this duration by monitoring
the listener’s tendency to stay tuned for only 20 minutes on
average each day and then jump someWhere else. Other
people With very short commutes Will seldom if ever skim
through very much buffer space (While long commuters Will
go through larger amounts). In addition, some people Will be
skimmers and some Won’t.
content every 20 minutes, the buffer should be about 25
a. Voice Mail
[0204] b. Cell phones
[0205] c. MP3 portable players
[0206] d. Radar detectors
e. MicroWaves
f. Cameras and camcorder
g. Clock radios
If a given station’s buffer isn’t used, some of that
station’s buffer memory Will be allocated to another use,
Which could be another station’s buffer or the jukebox.
The other principal competing use for buffer space
(in aggregate) is the jukebox. The system, then, Will allocate
memory betWeen the jukebox and the buffers. There Will be
a default buffer siZe (say, tWo hours per station and eight
hours in total). If the jukebox starts to impinge on that, the
system Will automatically let the jukebox take over “unused”
buffer space. When that is gone, the system Will announce
verbally that more jukebox space is needed and ask the user
if he or she Wants to delete songs or shrink the buffer siZes
[0210] h. Digital Watches
[0211] i. GPS navigation and positioning systems
[0218] When recordings are scheduled, the buffer memory
for each recording station’s audio track Will be automatically
expanded to accommodate the siZe of the recording. This
Will, hoWever, shorten the length of buffers for all stations.
[0212] It should be noted that none of these products has
the combination of computing poWer, poWer supply,
memory, or the robust speaker system is made available by
the recording/playback unit as illustrated in FIG. 1. As a
result, an improved user interface can be provided to these
connected devices.
[0213] The unit’s audio interface may be employed to
build in a Help Channel of audio information (With book
marks to make it navigable) stored behind the Options
buttons. Note that this Help channel could also store infor
mation about the car itself that the manufacturer might Wish
to have available to drivers on request. In other Words, our
audio system might be an infrastructure that the car OEM
might Wish to integrate With its system, supplying informa
The user Will be informed hoW a recording is impacting
available storage, letting him make a decision at that time
hoW to make any needed tradeoffs.
[0219] To save memory, the system Will automatically
categoriZe channels into talk or music stations. The confus
ing ones (like hip-hop) Will automatically be dumped into
the music class. Different encoding algorithms may be used
for music and talk stations or perhaps even for the talk
Within a music station. Some stations sWitch formats at set
times during the day, and this pattern may be used to reduce
the burden on the processor to discern betWeen talk and
music on the ?y.
Aug. 28, 2003
US 2003/0163823 A1
Another Way to conserve memory is to erase
[0236] HorsepoWer Will also be saved When time scaling
recorded content after it has been recorded, but before it has
been listened to, if it doesn’t ?t the user’s pro?le. There are
several examples of Where this could be used some of Which
assume the implementation of the previously discussed song
of talk stations is invoked as the decoding load is reduced as
identi?cation technology:
[0221] (1) The FFT algorithms could distinguish one
type of music from another and only keep the type
that ?t the listener’s observed or stated preference.
[0222] (2) The system could use ID technology to
identify commercials that have been listened to
before and delete them.
[0223] (3) Songs that have been saved in the jukebox
already could be deleted after identi?cation.
[0224] (4) Songs on the Never Again list could be
it’s being done that much faster.
[0237] With a high speed processor and an ef?cient coding
algorithm, encoding can be done faster than real time. In that
case, an encoder need be allocated for each station. The
content could be saved in un-encoded batches With the
encoded alternating the station being encoding. As a pref
erence setting, the time scaling feature could be automatic
for certain types, or for all types, of talk radio.
[0238] The user could also have the option of choosing the
level of compression desired and the quality of recording for
both music and talk in order to modulate both memory and
processing poWer usage.
[0239] If short buffers Were acceptable under certain cir
cumstances (for instance, When roving), or enough memory
deleted, or the recording stopped at a mid-point, as
Was available, encoding could be eliminated for some con
soon as the song Was identi?ed
tent With this unencoded content being stored for short
periods of time (until over-Written). If a section Were to be
saved to the Jukebox, it could be encoded at that time.
[0225] (5) Songs that have been played frequently in
the recent past could be erased. (Whereas a DJ
attempts to construct an optimal playlist based on the
assumption that you’ve heard all of it, this system
feature Would attempt to optimiZe the playlist based
on What you had actually heard recently.)
(6) Segments of talk shoWs that are continu
[0240] Potentially, once the system had identi?ed seg
ments as talk, or speci?c segments of talk (such an ad being
rerun), or speci?c songs, it could take the action of not
encoding or saving these portions, already saved songs,
songs that had been skipped over before, ads that had been
ally skipped by a speci?c listener (traf?c, World
listened to before, or other unWanted content.
neWs, etc.) could be deleted after recording if they
[0241] Alternatively, unWanted songs (as indicated from a
Never Again list, for instance), once identi?ed, could be
ignored and not run through the splitting process. Alterna
could be identi?ed via a metadata connection or
on-board content recognition.
[0227] Conserving Processing PoWer
[0228] The digital signal processor 103 performs multiple
tasks Which are computationally intensive, including
[0229] (1) Splitting (looking for music/talk bound
aries and other similar breakpoints)
Content (e.g. song) identi?cation
Encoding (e.g. MP3 compression)
Decoding (e.g. MP3 decompression
Time scaling
[0234] For music stations, the major horse poWer-saving
action Will be to minimiZe the combined load of splitting and
Song ID by minimiZing the amount of content that has to be
processed by each step. In addition, it Would conserve
processing poWer if talk content used a different encoding
algorithm. Song splitting can help ef?ciency by classifying
the content into talk or music. As a result, ID Work Will only
be done on music content, and the more ef?cient spoken
Word codec Will be used on the talk portions.
[0235] Other heuristic algorithms may be used to cut doWn
on poWer usage. For instance, by knoWing the genre of the
station, the usual playlist of the station being analyZed, and
even the list of recently played songs, the Song ID function
could look for the most probable matches ?rst. When
metadata of use in classifying a given audio track is to be
made available on a delayed basis, automatic bookmarking,
tively, only neW songs or songs on a Huntlist could be split
If much more content Was being saved than actu
ally being listened to, the system could save horsepoWer by
doing the splitting as needed, “on-demand”, so to speak.
That is When a listener sWitched to a channel, the content
being listened to could be decoded, split, and identi?ed (and
perhaps encoded again) before the user got to it. This
approach could be even more re?ned if the user Were sur?ng
from song to song via a method that landed him or her in the
middle of the song as opposed to the beginning. In this case,
Song ID and splitting Would only be done on demand—With
demand being indicated if the user listened to a certain
minimum length of the song or indicated a preference to go
back to the beginning to the song.
The system may impose the restriction that one or
more standard stations be a talk station in order to conserve
processing poWer.
[0244] No series of short cuts Will ensure that just the right
amount of processing poWer is alWays available. As such,
the system may employ an algorithm for deciding the
functions that do not get performed if there is not enough
horsepoWer. When memory is in surplus, some encoding can
be postponed and When that is not the case, then splitting and
Song ID could be postponed. Other measures could be to
stop buffering one or more channels, starting With the roving
station and then those standard stations that are listened to
voice/music recognition, other forms of computationally
burdensome analysis may be deferred until the metadata is
available to reduce the complexity of the analysis.
[0246] A tuner/encoder may be adapted to buffer only the
audio portions of broadcast TV signals. This Would provide
TV Tuner and other Broadcast Means
Aug. 28, 2003
US 2003/0163823 A1
signi?cantly more audio content for information-seeking
listeners. Metadata supplied by a server could be employed
to parse the captured audio stream to bookmark and describe
speci?c information for the listener.
[0247] The metadata for the TV audio, as Well as, any
metadata for radio stations could be embedded in a TV
a certain percent or absolute number of ads be listened to in
order to use the unit, record songs, or play them back
[0260] The use of an MP-3 CD-R or other removable and
Write-able media or a Wireless connection also opens up the
opportunity convey information back to the host computer,
broadcast signal. The data could alternatively be merged into
typically a PC but perhaps a server if a cellular Wireless
the audio stream With an IP-over—audio technology. An
alternative broadcast means to distribute metadata could be
to use a paging netWork and build a pager into each Radio.
connection is used. This back channel could be used for the
[0248] Business Models
[0249] The reception system that has been described can
form the basis for some unique business models, particularly
When combined With permission-based systems for using
various features of the system. The folloWing music-ori
ented features could be controlled by various “permissions”:
a. Recording a song
[0251] b. Not recording a song or other speci?c piece
of content on a selective basis because of a charac
teristic of that segment.
c. Splitting the song out With bookmarks
[0253] d. Saving the song (for a certain length of
e. Playing the song a certain number of times
[0255] f. Skipping all or part of an ad, or skipping a
[0256] g. Transferring the song from the system via
removable storage like ?ash memory or via a Wire
less connection. (This uploading could be done at
different quality levels depending on permissions.)
[0257] The permissions that could control access to these
features include the folloWing: Whether there is a Watermark
on the song indicating a particular copyright status; Whether
the song is ?agged as having a certain copyright status in the
Song ID database; or Whether the user already oWns the song
folloWing purposes:
[0261] (1) For e-commerce as listeners could indicate
the singles or CD, concert tickets, or other items that
they Were interested in purchasing as a result of
hearing them on the system.
[0262] (2) Listeners could request speci?c doWn
loaded information; such as samples off a CD they
may be considering purchasing.
[0263] (3) The listener could subscribe to a service
alloWing the removal off the system and/or retrans
mission of captured songs.
[0264] (4) Depending on the copyright rules of the
country in Which the system is used, it might be a
useful business model to have listeners pay a sub
scription fee to be able to record songs, or only
certain songs, using the system. This permission
might be granted only for a certain number of songs,
or songs that might only be played a certain number
of times, or that might only persist in memory for a
certain period of time.
[0265] (5) Information regarding listening activities,
habits and patterns could be uploaded for use in
monitoring the audience in an anonymous or aggre
gate fashion. This could generate information that
could be sold back to the content providers, radio
broadcasters, or record companies, or used by us to
improve the service. Information speci?c to the
listener that is conveyed to the host could be used
in another medium (eg a previously purchased CD);
[0266] i. Presenting better personaliZed ads
Whether the user subscribes to a particular on-line service
that Would normally provide access to that song; or Whether
the user subscribes to a “broadcast subscription service” that
Would give the listener certain rights to songs broadcast over
the air.
[0268] iii. Making guesses regarding What shoWs
[0258] The use of an MP-3 CD or other removable media,
a Palm Pilot linked to the unit’s USB port, or a Wireless
connection opens up the opportunity to interject not just
metadata into the system, but advertising as Well. The CD
could be used to transport personaliZed or generic ads to
users in conjunction With their doWnloading of metadata.
These ads could substitute or supplement any subscription
payments made in conjunction With the receiving of meta
data. The ads could be inserted into content (typically “White
space” that We create that is betWeen copyrighted Works) in
the folloWing Ways: at the beginning and end of the audio
track or at other major bookmarks; When the listener
sWitched channels; When the user sWitched modes (from
radio to CD, etc.); or betWeen songs, either in the jukebox
or the audio track.
[0259] An alternative business model, one that might
protect the interests of the broadcasters, Would mandate that
ii. Charging the correct fees for any sub
scription services
or stations a person might be interested in and
using that to do automatic recordings that the user
does not have to set up
iv. To be used in collaborative ?ltering
techniques that Will alloW one listener to bene?t
by knoWing What content other similar users lis
tened to. This could include the use of Hot Spots
as described in the above-noted patent applica
[0270] v. Explicit quality assessments by listeners
could be uploaded (user rated songs and commen
[0271] vi. Polls by the central server asking
Whether listeners agreed With a certain perspective
on a talk shoW.
[0272] An alternative to uploading this information to the
server, Would involve storing the information on the client
Aug. 28, 2003
US 2003/0163823 A1
Where it Would be used for the same purpose. For instance,
With some limited means to download data, a multiplicity of
ads could be delivered to the client, for instance via CD. The
client could then choose those that best ?t the user’s pro?le
based on listening data collected but stored on the client.
This approach Would Work Well With automatic recording as
Well. (Other functions mentioned above do require uploaded
information, such as proper billing.)
Note that all of these applications could apply to
both music or talk content.
[0274] Conclusion
[0275] The methods and apparatus Which have been
described are merely illustrative applications of the prin
ciples of the invention. Numerous modi?cations may be
made to the systems and components Which have been
described Without departing from the true spirit and scope of
the invention. For eXample, many of the techniques used for
selecting, recording and reproducing radio signals can be
used to advantage in television systems.
What is claimed is:
1. Aprogram reception and playback unit comprising, in
a digital memory for continuously and concurrently
recording a plurality of different broadcast program
a plurality of tuners for simultaneously receiving said
plurality of different broadcast program signals, each of
said tuners having a digital output for continuously
supplying a selected one of said broadcast program
signals in digital form to said digital memory, and
a player for accessing a selected one of said broadcast
program signals recorded in said digital memory at a
selected prior time.
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF