US 20030099345A1 (19) United States (12) Patent Application Publication (10) Pub. No.: US 2003/0099345 A1 Gartner et al. (43) Pub. Date: May 29, 2003 (54) TELEPHONE HAVING IMPROVED HANDS FREE OPERATION AUDIO QUALITY AND Publication Classi?cation METHOD OF OPERATION THEREOF (51) Int. Cl? . (52) Us. 01. ............................... .. 379/387.01; 379/388.01 H04M 1/00; H04M 9/00 (75) Inventors: Martin Gartner, Taufkirchen (DE); Thomas D. Slagle, Boca Raton, FL (57) (Us) Correspondence Address: Elsa Keller, Legal Assistant ABSTRACT A telephone having a hands-free mode of operation. The telephone includes a pair of microphones spaced apart from Intellectual Property Department each other. Each microphone receives sound in hands-free SIEMENS CORPORATION 186 Wood Avenue South mode of operation and provides audio signals representative of received sounds. The audio signals from each microphone may be converted to digital audio signals. The digital audio Iselin, NJ 08830 (US) signals are presented to a ?xed delay path and a variable (73) Assignee: Siemens Information (21) APPL NO; 09/994,405 ?ltered in an adjustable ?lter to remove noise based upon a prior determination of the noise source location and the (22) Filed; Nov, 27, 2001 voice spectrum derived from the digital audio signals. 102 I12 delay path. Audio signals from both paths are combined and 11B 1/14 / .__ ADC FIXED DELAY 128 / DIGITAL ADJUSTABLE AUDIO OUT V 126 * 104 118 / VARIABLE DELAY ADC 124 / 130 —~ 122 1/90 ‘?> PS - y d2 d ANALYSIS AND CONTROL no “ Patent Application Publication May 29, 2003 Sheet 1 0f 3 US 2003/0099345 A1 FIG. 1 \ /100 10s 3/104 106 FIG. 2 102 112 116 1/14 / ‘ FIXED ADC ’ DELAY 128 / 126 104 118 122 1/20 ADC . VARIABLE DELAY / y d 130 AUDIO 0m / 124 ‘?> PS ADJUSTABLE __ DEILGLILTEiLL ‘ DIGITAL d 2 ANALYSIS AND CONTROL 1 110 _ Patent Application Publication May 29, 2003 Sheet 2 0f 3 US 2003/0099345 A1 FIG. 3 142 144\ MICROPHONE SPACING NOISE BASELINE 145\ VOICE SPECTRUM 1“B\ EXTRACT DELAYS 150 \ 1 SET T2 NOISE BASELINE / 158 Patent Application Publication May 29, 2003 Sheet 3 0f 3 FIG. 4 , INCREASE /1482 US 2003/0099345 A1 May 29, 2003 US 2003/0099345 A1 TELEPHONE HAVING IMPROVED HANDS FREE OPERATION AUDIO QUALITY AND METHOD OF OPERATION THEREOF BACKGROUND OF THE INVENTION  1. Field of the Invention  The present invention is related to telephones and more particularly to telephones having a hands-free mode of BRIEF DESCRIPTION OF THE DRAWINGS  FIG. 1 shoWs an example of a preferred embodi ment telephone having a hands-free mode of operation;  FIG. 2 shoWs a preferred embodiment hands-free mode circuit for a speakerphone such as the telephone of FIG. 1; operation.  FIG. 3 is a How diagram shoWing steps to set up and use a preferred embodiment speakerphone;  2. Background   Typical state-of-the-art telephones often have a hands-free or speakerphone mode of operation, hereinafter generically “speakerphone.” Such a telephone may be located at a convenient location and placed in hands-free mode. Thereafter, speakers, e.g., teleconference participants, may remain stationary or move about Within range of the speakerphone as desired. The speakerphone microphone picks up all surrounding sound including background noise. This sound is transmitted to a listener at the other end of the call. Traditional speakerphones have a single microphone and are omnidirectional such that voice of the speaker and background noise are equally received and passed on to the listener.  Occasionally, background noise may be such that FIG. 4 is an example of how "52 is determined. DETAILED DESCRIPTION  FIG. 1 shoWs an example of a preferred embodi ment telephone 100 With a hands-free mode of operation that includes a ?rst microphone 102 and a second microphone 104 being used by a speaker 106 in the presence of a noise source 108. Preferably, the microphones 102, 104 are iden tical non-directional microphones and are mounted inter nally to the telephone 100 and spaced as far apart as the telephone casing alloWs, e.g., in the tWo front corners of the telephone casing. Thus, a sound from either of speaker 106 or noise source 108 arrives at each of the microphones 102, 104 at slightly different times, normally exhibited as phase differences. Thus, the dual microphone speakerphone exhib hands free operation is difficult to use if usable at all. Often its a directional microphone characteristic When the unde the background noise originates from a single source that layed signals from the microphones 102, 104 are combined. may be located at a ?xed location Within the room, e.g., from a noisy air conditioner or, from outside of the room such as  from street Work. To compensate for this background noise the microphone sensitivity may be loWered and the speakers may be requested to speak up. Sometimes this Works, sometimes it does not. Also, the noise may be such that setting the microphone sensitivity at one level is an unac ceptable solution, e.g., a pulsating type noise.  In an alternate embodiment the microphones are external to the speakerphone casing, Wired to the speaker phone. A larger distance betWeen the tWo microphones facilitates suppressing the loWer frequency noise sources. HoWever, this advantage is offset in that large spacing betWeen the tWo microphones 102, 104 may result in unequal signal volume betWeen the tWo microphones, espe cially, if the speaker is much closer to one microphone than Thus there is a need for a speakerphone With to the other. Accordingly, this alternate embodiment may capability of selectively removing background noise to provide improved audio quality, especially during hands free signal volume, e.g., one ampli?er, e.g., 118 as shoWn in FIG. operation. require additional logic/circuitry to compensate for different 2, may have an adjustable ampli?cation factor. SUMMARY OF THE INVENTION  Also, although the present invention is described noise ratio for telephones operating in hands free mode of herein as a digital embodiment, this is for example only. The hands-free telephone of the present invention may be imple operation; mented using analog components Without departing from the  It is a purpose of the invention to improve a signal  It is another purpose of the invention to improve the audio quality provided to a listener at a receiving ends of a hands free call;  The present invention is a telephone having a hands-free mode of operation. The telephone includes a pair of microphones spaced apart from each other. Each micro phone receives sound in hands-free mode of operation and provides audio signals representative of received sounds. The audio signals from each microphone may be converted to digital audio signals. The digital audio signals are pre sented to a ?xed delay path and a variable delay path. Audio signals from both paths are combined and ?ltered in an adjustable ?lter to remove noise based upon a prior deter mination of the noise source location and the voice spectrum derived from the digital audio signals.  Additional bene?ts and features of the invention Will be apparent from the folloWing detailed description taken together With the attached draWings. spirit or scope of the invention. Further, directional micro phones may be substituted for the above described non directional microphones 102, 104, provided they are directed toWards the expected speaker location and orthogo nal to the line de?ned by the microphones.  For purposes of description of the invention, the distance betWeen microphones 102 and 104 is referred to herein as x12. The distance betWeen speaker 106 and micro phone 102 is referred to herein as xul. The distance betWeen the speaker 106 and microphone 104 is referred to herein as xu2. The distance betWeen noise source 108 and microphone 102 is referred to herein as xnl. The distance betWeen noise source 108 and microphone 104 is referred to as xn2. Although, it is understood that the speed of sound varies With media and ambient conditions, for the purposes of this invention and, because normal operating conditions of a speakerphone for such a conference call are approximately constant, Thus, the the delay speed '51 betWeen of soundtheis tWo treated microphones as a constant is deter May 29, 2003 US 2003/0099345 A1 mined by x12 divided by c, i.e., t1=x12/c. Noise originating varies betWeen constructive and destructive interference, at noise source 108 in FIG. 1 arrives at microphones 102, 104 at times offset by (xn1—xn2)/c. Sound from a speaker 106 arrives at microphones 102, 104 at times offset by While the desired signals (xul, xu2) from the speaker or speakers alWays add constructively to provide a positive audio component. Taking the analog sound signal from (xu1—xu2)/c. microphones 102, 104 to be X1, X2, respectively, d1=X1  In the above alternate embodiment Wherein micro phones 102, 104 are external, '51 may be derived directly. A (i+'c1) and d2=X2(i), Where is the digital value of X at time i. Analysis and Control Unit 110 delays X2(i) betWeen 0 and 251, ?rst to identify the delay to minimiZe noise during tone may be radiated from one of the tWo microphones, e.g., 102. The delay betWeen When the tone originates at the ?rst microphone 102 and When it is received at the second microphone 104 is '51.  FIG. 2 shoWs a preferred embodiment hands-free mode circuit 110 for a speakerphone such as telephone 100 of FIG. 1. Sound signals from one microphone 102 pass through a ?xed delay path that includes an input ampli?er baseline determination and second to determine the delay to maximiZe xu/xn during voice spectrum analysis. Also, voice spectrum analysis results are applied to Adjustable Digital Filter 128 to enhance frequencies originating primarily from the speaker, and to dampen frequencies that originate pri marily or solely from the noise source 108. Therefore, as described hereinbeloW, each of these frequency bands are identi?ed in one of tWo different learning phases. In a ?rst 112, Analog-to-Digital Converter (ADC) 114 and ?xed delay 116. Coincidentally, sound signals from the second microphone 104 pass through a variable delay path that idle-state phase, the typical noise source spectrum is deter mined to identify the noise frequency bands. Then, in a includes an input ampli?er 118, an ADC 120 and an adjust mode and the composite sound that includes both noise and able variable delay 122. The outputs of ?xed delay 116 and the speaker’s voice is analyZed to determine the speaker’s variable delay 122 are combined in adder 126. The outputs of ADC 120 and ?xed delay 116 also are passed as inputs to Analysis and Control unit 124. The output of adder 126 is  Accordingly, having thus characteriZed the circuit passed to Adjustable Digital Filter 128. Analysis and Control unit 124 provides control for both adjustable variable delay 122 and Adjustable Digital Filter 128. Adjustable Digital Filter 128 provides a digital audio output that is the audio signal passed to a listener at the other end of the call. Phone status signals 130 are passed as inputs to Analysis and Control unit 124.  The ampli?ers 112, 118 of each path act as a speaker phase, the speakerphone is placed in hands-free frequency spectrum. response to both speaker input and noise input, the circuit may be calibrated to ?lter out noise. While it is preferred that the ampli?ers 112, 118 as Well as the ADCs 114, 120 are identical, in practice some slight differences alWays exist. These variations in or, differences betWeen components in each of the paths may be compensated, preferably, during factory calibration, e. g., by adjusting the ampli?cation factor of either or both of the ampli?ers 112, 118. By selectively adjusting variable delay 122 it is possible to folloW the preampli?er to amplify the sound signal from the particular connected microphone 102, 104. The output of ampli?ers speaker’s voice as the speaker moves about the set of 112, 118 are each passed to a respective ADC 114, 120. The ADCs 114, 120 convert the analog outputs from the corre directional microphone automatically to the user. As the variable delay 122 is changed to compensate or to coordi sponding ampli?ers 112, 118 to a digital output. The digital nate With changes of speaker location, background noise, Which originates elseWhere, is dampened or, possibly, removed. The degree of dampening for the background noise depends upon its angle of origin and Wavelength in output signal from ADC 114 is passed to a ?xed delay 116. Fixed delay 116 is set at '51 (i.e., x12/C). The digital output from ADC 120 is passed to adjustable variable delay 122. The Analysis and Control unit 124 may be a simple embed ded processor or microcontroller (not shoWn) and appropri ate program code, e.g., stored in a local read only memory (ROM) or electrically programmable ROM (EPROM). The Analysis and Control unit 124 controls delay in variable delay 122 and sets the ?lter bandWidth of Adjustable Digital Filter 128. Variable delay 122 has an adjustable delay of 62 that may be adjusted to values ranging betWeen 0 and 251.  In yet another alternate embodiment, both delays 116, 122 are adjustable variable delays, having a range betWeen 0 and '51. This alternate embodiment maintains overall circuit delay at a minimum. Accordingly, for this alternate embodiment, Analysis and Control unit 124 pro vides control to both adjustable delays.  Microphone input signals from microphone 102 microphones 102, 104. This is analogous to pointing a single relation to the noise source distance from the microphones 102, 104, i.e., loWer frequency sound (sub 100 HZ) tends to be non-directional. Since the loWer the frequency (f), the longer the Wavelength (8), loWer frequency sound is less subject to positional ?ltering and dampening. HoWever, such loW frequency noise may be removed With a simple loW pass ?lter or its equivalent in Adjustable Digital Filter 128.  FIG. 3 is a How diagram 140 shoWing set up and use of a preferred embodiment such as speakerphone 100 of FIG. 1. First, in step 142 the spacing betWeen the micro phones is input to determine "c1, e.g., entering the ?xed delay betWeen internal microphones 102, 104 at the factory or, for the above described external microphone embodiment, auto matically measuring the delay betWeen origination and reception of a tone. Then, in step 144 the background noise is checked. Typically, this check is done When the phone is (d1) and from microphone 104 (d2) are added constructively idle such as prior to making a call, at the beginning of a by setting "c2=t1—(xu2—xu1)/c, Which is maximum (261) conference call, etc. So, in this step 144 the phone is placed When the noise source is colinear With the microphones and separated from microphone 102 by microphone 104, i.e., in hands free mode and silence is maintained to generate a noise baseline With any noise sources that happen to be microphone 104 is betWeen noise source 108 and micro phone 102. Thus, for the above described range of '52, the signals at the tWo microphones 102, 104 may be added to produce a result Wherein the resulting noise component Within range of the phone. Next, in step 146, a second learning or voice baseline step, the speakerphone operates in hands-free mode and a speaker speaks from Within range of the phone to obtain a voice spectrum signal. The Analysis May 29, 2003 US 2003/0099345 A1 and Control unit 124 processes the signals from both micro phones to extract the voice spectrum from the background sounds using the background noise information obtained in step 144. The Adjustable Digital Filter 128 is adjusted to selectively enhance speech and suppress the background sounds.  So, in step 148 the Analysis and Control Unit 124 extracts delays both for noise sources and for voices as described hereinbeloW With reference to FIG. 4. In step 150, the optimum delay to maximiZe the voice to noise signal ratio (xu/xn) is set for '52, the adjustable variable delay 122 in the path from microphone 104. The path outputs from ?xed delay 116 and variable delay 122 are combined in adder 126 and that sum is passed to the adjustable digital ?lter 128. In step 152 the adjustable digital ?lter is adjusted to maximiZe speech and, simultaneously, suppress noise With the ?ltered result being passed to called parties. As long as the call continues in step 154 and While the speaker is speaking in step 156, this variable delay calibration may be repeated, periodically, in step 148 to folloW the speaker. Also, in step 156 When the Analysis and Control Unit 110 determines that no one is speaking, noise from the noise source may be re-analyZed in step 158 and the variable delay calibration repeated in step 148. When hands-free mode ends or the call ends in step 154, the ?ltering ends in step 160.  FIG. 4 shoWs an example of how "52 may be determined in step 148. Essentially, in each pass through step 148, "52 is varied slightly (slightly increased/decreased) and, then, the speaker’s voice to noise signal ratio (xu/xn) is checked until the optimum delay is found for '52, i.e., Where any change in "52 reduces xu/xn. Adjustable variable delay 122 is then set to the optimum value of "52 in step 150. During the initial pass through step 148, T2=T1 and xu/xn is marked or noted. Thereafter, in step 1482 the delay value for "52 is increased slightly and in step 1484, xu/xn is checked to determine if it has increased. If xu/xn increases in step 1484 an optimum value has not yet been identi?ed and, returning to step 1482, "52 is increased again. Iteratively increasing "c2 and checking xu/xn in steps 1482, 1484 continues until "52 is maximum (251) or, xu/xn is not found to have increased in step 1484. If xu/xn decreases after the ?rst increase of "52 in step 1482 xu/xn is not optimum. OtherWise When xu/xn decreases, the optimum value of xu/xn has been found in step 1486 (i.e., one increment beloW the current value) and in step 1488, "52 is backed off one increment (unless it is at its maximum value) and that value is passed to step 150.  If xu/xn decreases after the ?rst increase, then the optimum value for "52 has not been found in step 1486. So, the optimum value lies beloW the current value and in step 1490, the delay value for "52 is decreased slightly and in step 1492 xu/xn is checked to determine if it has increased. Steps 1490, 1492 are repeated iteratively, decreasing "c2 and check ing xu/xn until "52 is minimum (0) or xu/xn is not found to have increased in step 1492. Again in step 1488, "52 is backed off one increment (unless it is at its minimum value) and that value is passed to step 150.  Thus, the results of the analysis in the learning steps 144, 146 are combined to automatically maximiZe xu/xn and provide an optimal ?lter for the hands free phone. The result favors voice based signals over background noise.  Accordingly, the dual microphone hands free tele phone provides a microphone characteristic that is superior to single microphone telephones, While using a non-me chanical, dynamically adjustable reception direction. The background and voice analysis as described for FIG. 3 provides an optimal ?lter for the dual microphone telephone. In particular analysis is simple enough that recalibration may be done periodically, manually or automatically throughout the call to identify background noise. The back ground noise may be analyZed While the telephone is idle or during hands free operation, if no one is speaking. The digital audio output may be provided to any typical tele phone equipment, e.g., converting the ?ltered digital audio back to an analog signal for analog transmission or, sending it as voice over internet protocol (VoIP).  Thus, the dual microphone telephone of the present invention provides a signi?cant audio quality improvement during hands free operation over prior art bands free tele phones. Further, automatic recalibration may not require users to perform additional tasks or, at most, may require performing minimal additional tasks, e.g., initiating each of the learning steps.  While the invention has been described in terms of preferred embodiments, those skilled in the art Will recog niZe that the invention can be practiced With modi?cation Within the spirit and scope of the appended claims. What is claimed is: 1. A telephone having a hands-free mode of operation, said telephone comprising: a ?rst microphone receiving sound in hands-free mode, and providing ?rst audio signals representative of received sounds to a ?rst delay path; a second microphone receiving said sounds in hands-free mode and providing second audio signals representa tive of said received sounds to a second delay path, said second microphone spaced a selected distance from said ?rst microphone; an adder combining said ?rst audio signals from said ?rst delay path With said second audio signals from said second delay path; an analysis and control unit analyZing received said signals and adjusting delay through said second delay path; and an adjustable ?lter receiving combined said signals from said adder and ?ltering noise from said combined signals. 2. A telephone as in claim 1 Wherein said ?rst delay path is a ?xed delay path, said second delay path is a variable delay path and said ?rst audio signals from said ?xed delay path and said second audio signals from said variable delay path are digital audio signals. 3. A telephone as in claim 2 Wherein said ?xed delay path provides a delay proportional to the selected distance betWeen said ?rst microphone and said second microphone. 4. The telephone as in claim 2 Wherein the variable delay inserts a delay having a range less than tWice the delay of said ?xed delay path. 5. A telephone as in claim 1 Wherein said adjustable ?lter is an adjustable digital ?lter providing a digital audio output. 6. A telephone as in claim 2 Wherein each of said ?xed delay path and said variable delay path comprises: May 29, 2003 US 2003/0099345 A1 an ampli?er receiving an analog signal from a connected microphone; and an analog-to-digital converter (ADC) converting an out put of said ampli?er to a corresponding digital signal. 7. A telephone as in claim 6 Wherein said digital signal from said ADC in said variable delay path is provided to said analysis and control unit. 8. Atelephone as in claim 2 Wherein said output from said ?xed delay path is provided to said analysis and control unit. 9. A telephone as in claim 2 Wherein said analysis and control unit further sets ?lter values in said adjustable digital ?lter. 10. A telephone as in claim 2 Wherein the analysis and control unit comprises: means for varying the delay of said adjustable delay path; b) taking a voice baseline at each of said microphones, said voice baseline providing a voice frequency spec trum of a speaker’s voice; c) comparing said voice baseline With said noise baseline to determine a substantially optimum delay for a signal path from one of said microphones; d) setting a delay in said signal path responsive to said optimum delay; and e) ?ltering noise associated With said noise spectrum. 14. A method as in claim 13 Wherein steps c) through e) are periodically repeated throughout a hands-free call. 15. Amethod as in claim 14 Wherein at least one idle time is identi?ed in said hands-free call and, at each said at least one idle time a neW noise baseline is extracted from signals determining means for determining a ratio of a voice signal to a background noise signal; and means for identifying an increase in said ratio responsive to delay changes in said adjustable delay path. 11. A telephone as in claim 10 Wherein said analysis and control unit further comprises: means for extracting a noise spectrum from a ?rst signal; and means for extracting a voice spectrum from a composite signal, extracted said voice spectrum being compared from said microphones. 16. A method as in claim 13 Wherein the step c) of comparing said voice baseline With said noise baseline comprises the steps of: i) incrementally increasing said delay; ii) comparing a voice to noise signal ratio at said increased delay With a previous voice to noise signal ratio to determine if said voice to noise signal ratio is increased; and, against said noise spectrum in said determining means. 12. A telephone as in claim 11 Wherein said adjustable iii) repeating steps i) and ii) until said voice to noise signal digital ?lter is an adjustable bandpass ?lter and said analysis and control unit adjusts said adjustable bandpass ?lter to remove signals having frequencies outside of said extracted voice spectrum. 17. A method as in claim 13 Wherein the step c) of comparing said voice baseline With said noise baseline 13. A method of controlling a speakerphone, said speak erphone having at least tWo microphones spaced a selected distance from each other, sound signals from each of said microphones being combined in said speaker phone and presented as a voice output from said speakerphone to a party at another end of a hands-free call, said method comprising the steps of: a) taking a noise baseline at each of said microphones, said noise baseline providing a noise frequency spec trum of background noise; ratio is determined not to have increased. comprises the steps of: i) incrementally decreasing said delay; ii) comparing a voice to noise signal ratio at said increased delay With a previous voice to noise signal ratio to determine if said voice to noise signal ratio is increased; and, iii) repeating steps i) and ii) until said voice to noise signal ratio is determined not to have increased.
* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project