Begault, Brustad and Stanley
Tape analysis and authentication
TAPE ANALYSIS AND AUTHENTICATION
USING MULTI-TRACK RECORDERS
DURAND R. BEGAULT, BRIAN M. BRUSTAD AND ANDREW M. STANLEY
Audio Forensic Center, Charles M. Salter Associates
130 Sutter St., Suite 500, San Francisco CA 94104
Durand.Begault@cmsalter.com
www.audioforensics.com
Techniques and advantages of multi-track (> quarter track) playback heads for forensic analysis of standard
monophonic and stereophonic cassette and micro-cassette tape recordings is discussed. The time-domain waveform for
recorded signatures can be analyzed in terms of relative timing offset for determining azimuth of the record head used
in making a specimen tape. Additionally, excursion of the erase signature into the guard band region is a reliable
indicator of an original versus a copied erase signature. Comparative advantages over magnetic development techniques
are discussed.
1 INTRODUCTION;
GENERAL
AUTHENTICATION
IN
In many cases, an audio forensic expert is called upon to
examine taped evidence to provide an opinion on
whether or not a tape has been “edited” or “doctored” in
any way. Specifically, this translates into an analysis of
the temporal sequence of events found on the tape that
correspond to record start, pause, and stop operations of
one or more tape recording devices. This typically
includes the analysis of “record event signatures”
corresponding to the interaction of the tape surface with
the electrical activation and deactivation of AC-bias
record and erase heads, and/or contact with a permanent
magnet erase head.
Record event signatures, if they exist, may or may not
be revealed via a number of techniques that are applied
in the audio forensic laboratory. This includes aural
analysis (“critical listening”); visual analysis of the
time-domain waveform (“waveform analysis”); and via
visual analysis of the bitter pattern of the magnetized
surface of the tape (“magnetic development”). The bitter
pattern, a visual representation of the magnetized
portions of the tape rendered visible through a lowpower microscope, results from the application of
micron-sized iron particles in a liquid suspension to the
tape known as “ferrofluid”. A camera or video recorder
can be used to document the event. Several key papers
in the literature address the use of these techniques
[1,2,3]. Figure 1 illustrates a comparison of the same
record event seen via Magnetic development (top) and
time-domain waveform (bottom).
RECORD HEAD……………………..ERASE HEAD
Figure 1: Magnetic development (top) and timedomain waveform (bottom) of the same record
event (record stop signature). The magnetic
development photo has been reversed to match the
sequence of events as would be transmitted to the
computer by the playback head.
In addition to examining record event signatures to
determine continuity or editing, audio forensic experts
are also sometimes asked to confirm if an evidence tape
is an original recording, or a copy. Evidence of an
original versus a copied tape usually results from similar
record event signature analysis, whenever possible.
While it is possible to determine that a record stop event
is a copy via the analysis techniques just described, an
audio forensic expert can never determine with absolute
certainty if a recording is truly “original”; one can only
state that the recording is consistent with an original
AES 26th International Conference, Denver, Colorado, USA, 2005 July 7–9
1
Begault, Brustad and Stanley
Tape analysis and authentication
Figure 2. Tape head configuration for ¼ track stereo tape (indicated as C, boxed). The two bottommost tracks
correspond to left and right channels of the “A” side of a tape; the two uppermost tracks correspond to the left
and right channels of the “B” side of the tape. The middle area between the A and B side head pairs is referred
to as the center ‘guard band’.
recording. As an example, an original recording can be
recorded to a computer, digitally edited, and then played
back to a cassette recorder containing a tape with
“leader” at the start and end. The leader cannot be
recorded upon since it has no metallic surface that can
be magnetized. The action of the tape recorder
corresponding to the record start and record stop events
cannot be observed via either waveform or magnetic
development analysis.
Another area of authentication not considered here is
matching a tape to a specific tape recorder used to make
the evidence recording. It is on occasion useful for an
attorney to have an audio forensic expert impugn the
reliability of a witness’s testimony in cases where a
specific tape recorder is claimed to have been used to
make a specific evidence tape. Intra-machine variability
viewed with inter-machine variability makes it far easier
for an expert to eliminate a specific machine rather to
identify it. (There is a corresponding observation
regarding voice identification; see paper by F. Poza and
D. Begault, Voice identification and elimination using
aural-spectrographic protocols, this conference).
Finally, the forensic audio expert and legal practitioners
must bear in mind the difference between what has been
termed “Technical Authentication” versus “Legal
Authentication”. While related, the audio forensic
expert is not required nor are they responsible to legally
authenticate evidence in the legal sense [1]. Technical
authentication is within the expertise of the audio
forensic expert.
2 WAVEFORM ANALYSIS
Waveform analysis begins with the archival playback of
the evidence tape into a computer for digital storage,
using a high quality tape recorder for playback having
the same type of head configuration used in the original
recording. For example, a ¼ track stereo cassette tape,
which is one of the most popular formats, is played back
for digital storage and analysis on a computer using a ¼
track stereo cassette tape machine (see Figure 2 for
various formats of tape). Procedures such as those
outlined in Audio Engineering Standard 43-2000
“Criteria for the authentication of analog audio tape
recordings” can be followed.
A computer with calibrated, high-quality analog-digital
converters and appropriate software essentially act as a
high-quality tape recorder. Once archived, the use of an
audio recording waveform editor greatly simplifies
navigation throughout the ‘unaltered’ (direct, one-one)
copy of the audio contained on the entire evidence tape,
and critical listening can be combined with visual
observation to mark significant events for further
inspection. Similar or more detailed visual analyses
corresponding to a photograph of an oscillograph trace
can be accomplished by “zooming” in on the display,
including with “scientific analysis” software (such as
Mathworks’ MATLAB).
Spectrographic analysis, which analyzes the strength of
various frequency components, can also be used “on the
fly”. Record event signatures, particularly pause
signatures, are also sometimes detectable in this
manner. (A spectrum is analogous to the bands of light
seen when a light source is placed in front of a prism.
Spectrographic and other frequency analysis techniques,
such as Fourier analysis, are beyond the scope of this
discussion).
Of the various types of record event signatures, record
stops are most visible using waveform analysis due to
the characteristic large peak of the record head electrical
discharge, followed by the low-frequency erase head
discharge (see Figure 1). The erase head signature is
also commonly audible as low-frequency “thump”.
Record start signatures are sometimes more difficult to
see, as one observes an energizing of the waveform as
AES 26th International Conference, Denver, Colorado, USA, 2005 July 7–9
2
Begault, Brustad and Stanley
an amplitude increase, often convolved with the
waveform of the input [2]. Pause signatures are
sometimes, but not always easy to observe using just
waveform analysis. Overall, waveform analysis allows a
relatively simple means for examining discontinuities
across long recordings, which can be observable from
either the presence of signatures, just discussed, and/or
from changes in background noise.
3
Tape analysis and authentication
magnetic development. A dubbed version of this erase
head signature would be the same height as the record
head. This is due to the fact that erase heads are
generally manufactured to exceed the vertical extent of
the record head, to insure complete erasure of
recordings within a range of record head alignments.
MAGNETIC DEVELOPMENT
For short-duration events, magnetic development has
historically been considered to be superior to waveform
development for forensic analysis. For example, one can
immediately observe the track width and configuration
from a photograph of the bitter pattern, as well as
determine recorded from unrecorded tape. Comparing
Figure 1 to Figure 2 shows that the examined tape in
Figure 1 is unrecorded on side B and that the track
configuration of side A corresponds to a ¼ track tape
configuration.
Magnetic development also clearly indicates in some
instances an original versus a copied record head
signature. Figure 3, left shows the non-flat surface of a
record head signature from an original recording, caused
by vertical motion (droop) of the tape resulting from the
mechanical action of the recorder. Figure 3, right shows
the same record head signature on a copied tape. There
is no droop on the copied tape; however, this signature
could have also been found on an original tape, caused
by a recorder that did not move the tape vertically.
Examination of the erase head mark would provide
additional clarification.
Figure 3. Magnetic development of record head
signatures. Left: from an original recording.
Right: from a copy of the same tape.
Less ambiguously, magnetic development can also
reveal over-recordings (two overlapping bitter patterns
or via two record head edge traces). The difference in
height of the two traces in Figure 4 indicates two
different recordings, one caused by an over-recording.
Also note the fact that the vertical height of the erase
head signature is greater than the vertical height of the
record head trace. This is an illustration of the means by
which an original erase head signature is visible via
Figure 4. Magnetic development of a recordover signature.
4 DISADVANTAGES
&
CAUTIONS
MAGNETIC DEVELOPMENT
RE
Magnetic development has advantages over waveform
analysis or other methods for detecting and observing
the types of signatures just discussed. However, there
are also several disadvantages to the method that must
be considered. First, the process is time-consuming and
consequently expensive for clients. This is because only
a small portion of the tape, maybe several inches, can be
removed safely from the tape at one time.
Consider that at a speed of 1 7/8 inches per second,
twelve inches of tape corresponds to about six seconds.
This is fine for investigating a single event, such as a
record stop where the distance between the record and
erase head is usually about one inch. But an examiner
cannot practically magnetically develop an entire tape; a
30minute tape would require development of 3375
inches of tape and investigation of about half as many
photographs. Hence, magnetic development is typically
performed to confirm the results of critical listening
and waveform analysis for specific events.
Another disadvantage of magnetic development relative
to waveform analysis is that the signals must be
relatively high amplitude in order to be visible.
“…certain record and erase head events from some tape
recorders may be too weak to be exhibited” [1]. In fact,
record events such as record stops that are visible with
magnetic development are nearly always audible. By
contrast, waveform analysis allows investigation of
signals that exceed the (usually very low) noise floor of
the computer recording. (In a future study, we plan to
compare the minimum detectable signal level
observable using magnetic development).
AES 26th International Conference, Denver, Colorado, USA, 2005 July 7–9
3
Begault, Brustad and Stanley
Tape analysis and authentication
Figure 5. Signal loss due to magnetic development of tape having a pink-noise spectrum. Left: result of subtracting the
“before” from the “after: spectra for six tapes. Right: same data with 2nd order polynomial curve fit.
However, the principal disadvantage of magnetic
development is potential damage to the evidence tape
itself. Magnetic tape used in cassettes consists of an
extremely thin base film (6 – 12 µm) with an oxide
coating of about 5 µm, and is susceptible to stretching
and folding [4]. Its preservation is by design dependent
on the protective cassette housing and reels, which only
exposes a small portion of the tape at one time while
keeping a flat, tight wind on each tape hub.
Figure 5 shows results of tests from the Audio Forensic
Center laboratory using specimen cassette tapes to test
for signal loss from magnetic development conducted
by two experienced examiners. Six virgin cassette tapes
(Maxell XLII-S 90) had a recording of pink noise
applied using a Sony TCD-5 professional cassette
recorder. The “before” spectrum was analyzed for each
tape individually, using a Hewlett-Packard HP5670A
real-time FFT analyzer (100 averages of a four-second
looped segment, Hann window). Then, a small amount
of ferrofluid (“Kyread F concentrate” manufactured by
Kyros Corporation, in a 1,1,2 Tricholotrifluoro-ethane
(“Freon”) base) was applied to each tape and allowed to
dry. The residue was wiped off using a soft disposable
cloth (“Kimwipe”) and then the tape was rewound and
re-played for FFT analysis using the same cassette
playback device to obtain an “after” spectrum in the
same manner as the “before” spectrum. The tape
playback head was cleaned and dried before each
playback to control for residue from previous tapes.
Additional tests conducted in our laboratory using
individual sine wave frequencies also indicated signal
loss consistent with the results of Figure 5. Table I
shows the mean value and raw data for three tapes
prepared with three sine wave tones (1.7, 4 and 10 kHz)
for the “before” and “after” spectra. Similar or greater
losses were observed using a ferrofluid manufactured by
FerroTec (type EMG 408). Additional ‘wiping’ of tapes
did not offset the observed losses.
Taken together, these data indicate that magnetic
development of tapes can cause damage to an evidence
tape in terms of signal loss, particularly in higher
frequencies. The amount of loss may be due to the
amount of iron particles not removed during a specific
test. This loss may impact alternative methods of
examining a tape after magnetic development is
performed, including waveform analysis, spectral
analysis and critical listening, and may also impact the
enhancement of speech recordings. Furthermore, the
signal loss is probably permanent, due to the interaction
of the metal particles of the ferrofluid with the
magnetized oxide surface. Finally, there is the
possibility of folding and creasing of the tape due to its
removal from its protective shell, which can
permanently damage the high frequency content of a
tape.
1
2
3
MEAN
1.7 kHz
1
2
3
2.0
4 kHz
1
3
3
2.3
10 kHz
3
4
2
3.0
Table I. Signal loss (dB) from magnetic
development of three tapes recorded with 1.7
kHz, 4 kHz and 10 kHz tones (Sony TCD5),
along with mean values.
AES 26th International Conference, Denver, Colorado, USA, 2005 July 7–9
4
Begault, Brustad and Stanley
5 MULTI-TRACK
ANALYIS
HEAD
Tape analysis and authentication
WAVEFORM
It would be advantageous to utilize waveform analysis
as a substitute for magnetic development in order to
prevent damage to an evidence tape. The technique of
using multi-track playback heads for forensic analysis
of standard monophonic and stereophonic cassette and
micro-cassette tape recordings is a useful alternative for
determining whether record stop signatures were made
on the evidence tape (‘original record stop signature’) or
on a later generation of the tape (‘copied record stop
signature’). This is a form of technical authentication
that can be useful in making probative statements
regarding the originality of the evidence.
Audio Forensic Center has evaluated multi-track
cassette units from Tascam Corporation 688 (8 track
standard Cassette) and JBR Technology (4 track
microcassette, and 2 track-9 track magneto-resistive
head for standard Cassette) for their application to
determining copied from original tapes, and for other
functions. Preliminarily, it is possible to report that the
multi-track units reveal pertinent information regarding
erase head marks and track configuration that would
otherwise only be observable via the bitter pattern
produced by magnetic development.
The 8-track head configuration labeled “B” in Figure 2
is that used in the Tascam multitrack recorder 688.
Figure 6 shows the eight tracks via magnetic
development of the bitter pattern from recording of a
315 Hz sine wave. The middle two tracks (numbering
sequentially, tracks 4 and 5) are indicated with arrows in
Figure 6. The location of these tracks lies within the
guard band area of ½ track and ¼ track recordings (see
configurations “C” and “D” in Figure 2), and erase head
signatures typically only occur in these track regions
with original tapes. However, due to head misalignment
on original recordings, the middle tracks are sometimes
not completely centered over the guard band.
Figure 6. Magnetic development showing layout of
tracks (dark stripes) on Tascam 688 8-track
cassette playback head. Arrows indicate tracks that
correspond to the center guard band of a standard
¼ track and ½ track recording.
Figure 7. Layout of head on JBR Technology’s 2
track-9 track magneto-resistive head cassette
playback machine (approximate-not to exact scale).
Left (open rectangles): location of stereo ¼ track
record-playback head (width 600 µm); center (solid
rectangles): location of magnetoresistive head
(width 70 µm); right: approximate erase head
location.
Figure 6 shows how the arrangement of the track heads
for the 2 track-9 track magneto-resistive head cassette
playback machine corresponds to the standard head
layout of a ¼-track stereo cassette. As shown in Figure
1 and detailed in Figure 7, a tape recorder erase head is
wider than the width corresponding to the audio tracks.
By using a multitrack magnetoresistive head with
narrow gap width (70 µm) there can be a more precise
playback head track corresponding to the guard band
region between the “A” and “B” side of the tape.
Waveform analysis of the output of track 1 of this head
will show detailed time-domain information from the
erase head, if the signature was made on the original
tape. The relative attenuation of audio in track 5 is an
indicator of whether the analyzed tape wave ¼ track
stereo or ½ track monaural.
Consider the simple example of a tape where the
recording is stopped perhaps one minute into the
recording. On the original tape, a record stop signature
will be visible via magnetic development with the erase
head extending beyond the top of the audio track and
into the “guard band” region between sides. On a copy
the erase head mark would not be visible on track 1 (and
sometimes, track 2).
AES 26th International Conference, Denver, Colorado, USA, 2005 July 7–9
5
Begault, Brustad and Stanley
ORIGINAL
Tape analysis and authentication
COPY
Figure 8. Time domain view of erase head
signatures as reproduced by the magnetoresistive
head for ½ track original (left side) and ¼ track
copy (right side) . Top: track 1; Bottom: track 2.
Figure 8 shows a comparison between an original tape
(1/2-track hand-held Radio Shack recorder) and a dub
made of the same tape between two ¼ track recorders.
Figure 9 shows spectrograms of the same data from 0500 Hz. The relatively higher frequencies seen in the
erase head for Track 1 on the original tape are missing
from Track 2.
6
ANALYSIS OF AZIMUTH ERROR
Azimuth refers to the position and angle of the record
head or playback head in relation to an audiotape.
Azimuth error results if a discrepancy exists between
the azimuth of the record head and the playback head.
Recorded audio data regarding the alignment of the
azimuth of a record head may be used as an identifier to
a particular recorder.
A reduced level of signal and signal degradation can
occur if a misalignment of the record head causes an
azimuth error during playback. Audio enhancement
may require a playback head be aligned to the azimuth
of a record head in order to improve signal quality from
an evidence tape.
Figure 9. Spectogram of erase head signature, 0500 Hz. Top: tracks 1,2 for original tape. Bottom:
tracks 1,2 for dub.
The azimuth error resulting from an improperly aligned
record head of a magnetic tape recorder can be
quantified by analyzing multi-channel playback in the
time domain. The time of arrival of the peak excursion
of a record signal can be easily compared across a
multitrack time domain display. Azimuth errors will be
exhibited as successively increasing delta time delay
values over adjacent heads of a multitrack playback
recording. The delta time delay can then be used to
determine the degree offset, top-to-bottom, of the
azimuth.
AES 26th International Conference, Denver, Colorado, USA, 2005 July 7–9
6
Begault, Brustad and Stanley
A simplified value for degree of azimuth error is
represented as:
θ = tan-1 ( Δt * splayback )
( Δh)
Tape analysis and authentication
REFERENCES
[1]
J. Gruber, F. Poza, and A. Pellicano, “Audiotape
Recordings: Evidence, Experts and Technology”
American Jurisprudence. Trials. vol. 48,
Lawyers Cooperative Publishing (1993).
[2]
B. Koening, “Authentication of Forensic Audio
Recordings” Journal of the Audio Engineering
Society vol. 38, no. 1/2 (1990).
[3]
T. Owen, “Forensic Audio and Video-Theory
and Applications” Journal of the Audio
Engineering Society vol. 36, no. 1/2 (1988).
[4]
C. D. Mee and E. D. Daniel (ed.), Magnetic
Recording
Handbook.
Technology
and
Applications, McGraw-Hill (1990).
Where:
θ = Degree of azimuth error
Δt = Time delay between heads in seconds
splayback = Speed of tape playback in cm/s
Δh = distance between heads in cm
This equation assumes the multitrack head used for
forensic analysis has no azimuth error. Any error in the
azimuth of the multitrack head would exaggerate the
negative or positive delta time delay values over
adjacent poles.
The precision of the multitrack head in determining
angle of azimuth of an evidence recording is dictated by
the sampling frequency of the digital copy of the
recording (resolution of Δt) and the speed of the tape
during playback (Splayback). The 9 track magnetoresistive head yields a precision of 0.04 degrees at a
sampling rate of 44.1 khz and a tape speed of at 4.76
cm/s. Higher sampling rates are easily obtained using
high-quality ‘off-the-shelf’ analog-digital converters.
7
CONCLUSION
While not a replacement for magnetic development in
some circumstances, the advantages of multi-track
waveform capture for forensic analysis of standard
monophonic and stereophonic analog tape are
numerous. Multi-track waveform analysis provides far
more insight into the detail of an evidence tape than
waveform analysis of one or two tracks that correspond
to the original recorded track width. The tape may be
examined for many important features that previously
could only be examined using magnetic development.
Two features of multi-track waveform analysis were
reviewed. First, the excursion of the erase signature into
the guard band region is a reliable indicator of an
original versus a copied erase signature, particularly
when using 9 narrow tracks extending over half the
vertical height of the tape. Second, the time-domain
waveform for recorded signatures can be analyzed in
terms of relative timing offset for determining the
azimuth of the record head of a specific tape recorder
used in making an evidence tape.
AES 26th International Conference, Denver, Colorado, USA, 2005 July 7–9
7
Download PDF