PCT
`INTERNATIONAL APPLICATION PUBLISHED UNDER THE PATENT COOPERATION TREATY (PCT)
`WO 00/21194
`(51) International Patent Classification 6 :
`H03G 3/20, H04B 15/00, H04R 25/00
`
`WORLD INTELLECTUAL PROPERTY ORGANIZATION
`International Bureau
`
`(11) International Publication Number:
`
`Al
`
`(43) International Publication Date:
`
`13 April 2000 (13.04.00)
`
`(21) International Application Number:
`
`PCT/US99/23234
`
`(22) International Filing Date:
`
`5 October 1999 (05.10.99)
`
`(81) Designated States: AU, CA, JP, European patent (AT, BE,
`CH, CY, DE, DK, ES, FI, FR, GB, GR, IE, IT, LU, MC,
`NL, PT, SE).
`
`(30) Priority Data:
`09/168,101
`
`8 October 1998 (08.10.98)
`
`Published
`With international search report.
`
`US
`
`(71) Applicant: RESOUND CORPORATION [US/US]; Seaport
`Centre, 220 Saginaw Drive, Redwood City, CA 94063 (US).
`
`(72) Inventors: PUTHUFF, Steven, H.; 13001 Saratoga-Sunnyvale
`Road, Saratoga, CA 95070 (US). T AENZER, Jon, C.; 888
`Laverne Way, Los Altos, CA 94022-1107 (US). GEDDES,
`Earl, R.; 1388 Medinah Drive, Itasca, IL 60143 (US).
`
`(74) Agent: KREBS, Robert, E.; Bums, Doane, Swecker & Mathis,
`LLP, P.O. Box 1404, Alexandria, VA 22313-1404 (US).
`
`(54) Title: DUAL-SENSOR VOICE TRANSMISSION SYSTEM
`
`PD Mic
`
`PK.DET.&
`THRESH.
`
`,,
`
`~
`
`(0
`
`a,.o
`
`(57) Abstract
`
`A pickup system (10) utilizes both sound pressure disturbances sensed by a microphone (20) and bone conduction sensed by a
`vibration sensor (30) such as an accelerometer, velocity sensor or displacement sensor to pick up and faithfully transmit a talker's voice
`despite acoustically noisy ambient environments. The system senses the microphone signal amplitude and adjusts the gain of associated
`amplification circuitry (23, 26) to effectively and smoothly fade in the microphone signal and fade out the vibration sensor signal as ambient
`noise increases. The pickup system (10) may be part of a stand alone ear level communication transmitter worn behind the talker's ear, or
`it may be implemented in a telephone system or other communication systems.
`
`- i -
`
`Amazon v. Jawbone
`U.S. Patent 8,019,091
`Amazon Ex. 1004
`
`

`

`FOR THE PURPOSES OF INFORMATION ONLY
`
`Codes used to identify States party to the PCT on the front pages of pamphlets publishing international applications under the PCT.
`
`AL
`AM
`AT
`AU
`AZ
`BA
`BB
`BE
`BF
`BG
`BJ
`BR
`BY
`CA
`CF
`CG
`CH
`CI
`CM
`CN
`cu
`CZ
`DE
`DK
`EE
`
`Albania
`Annenia
`Austria
`Australia
`Azerbaijan
`Bosnia and Herzegovina
`Barbados
`Belgium
`Burkina Faso
`Bulgaria
`Benin
`Brazil
`Belarus
`Canada
`Central African Republic
`Congo
`Switzerland
`COte d'Ivoire
`Cameroon
`China
`Cuba
`Czech Republic
`Gennany
`Denmark
`Estonia
`
`ES
`Fl
`FR
`GA
`GB
`GE
`GH
`GN
`GR
`HU
`IE
`IL
`IS
`IT
`JP
`KE
`KG
`KP
`
`KR
`KZ
`LC
`LI
`LK
`LR
`
`Spain
`Finland
`France
`Gabon
`United Kingdom
`Georgia
`Ghana
`Guinea
`Greece
`Hungary
`Ireland
`Israel
`Iceland
`Italy
`Japan
`Kenya
`Kyrgyzstan
`Democratic People's
`Republic of Korea
`Republic of Korea
`Kazakstan
`Saint Lucia
`Liechtenstein
`Sri Lanka
`Liberia
`
`LS
`LT
`LU
`LV
`MC
`MD
`MG
`MK
`
`ML
`MN
`MR
`MW
`MX
`NE
`NL
`NO
`NZ
`PL
`PT
`RO
`RU
`SD
`SE
`SG
`
`Lesotho
`Lithuania
`Luxembourg
`Latvia
`Monaco
`Republic of Moldova
`Madagascar
`The fonner Yugoslav
`Republic of Macedonia
`Mali
`Mongolia
`Mauritania
`Malawi
`Mexico
`Niger
`Netherlands
`Norway
`New Zealand
`Poland
`Portugal
`Romania
`Russian Federation
`Sudan
`Sweden
`Singapore
`
`SI
`SK
`SN
`sz
`TD
`TG
`TJ
`TM
`TR
`TT
`UA
`UG
`us
`uz
`VN
`YU
`zw
`
`Slovenia
`Slovakia
`Senegal
`Swaziland
`Chad
`Togo
`Tajikistan
`Turkmenistan
`Turkey
`Trinidad and Tobago
`Ukraine
`Uganda
`United States of America
`Uzbekistan
`Viet Nam
`Yugoslavia
`Zimbabwe
`
`- ii -
`
`

`

`WO 00/21194
`
`PCT /US99/23234
`
`DUAL-SENSOR VOICE TRANSMISSION SYSTEM
`
`BACKGROUND OF THE INVENTION
`
`FIELD OF THE INVENTION
`
`The invention relates to two-way communication devices, and more
`
`5
`
`particularly, to ear level voice pickups used in two-way communications.
`
`DESCRIPTION OF RELATED ART
`
`Existing ear level pickup technology uses a conventional microphone to
`
`sense and relay the wearer's voice to a listener at a remote location. The
`
`microphone responds to sound signal air pressure disturbances and converts these
`
`10
`
`to a representative electrical signal which is then provided to a transmitter. The
`
`transmitter relays the electrical signal, via any appropriate means such as radio
`
`waves or telephone lines, to a receiver at a remote location, where the electrical
`
`signal is reconverted, using a loud speaker, to an audible sound signal for hearing
`
`by the listener at the remote location.
`
`15
`
`A problem afflicting such systems is their susceptibility to degradation by
`
`high ambient noise levels present in the environment. When the pickup wearer is
`
`immersed in a noisy environment, the environmental sounds interfere with the
`
`talker's voice and, above certain levels, overshadow that voice and frustrate
`
`intelligibility at the remote location. As these systems are often targeted for
`
`20
`
`wearers whose lines of work necessitate exposure to high ambient noise levels--
`
`e.g., government field agents such as secret service employees assigned to
`
`- 1 -
`
`

`

`WO 00/21194
`
`PCT/US99/23234
`
`-2-
`
`protect officials and to communicate in the midst of large crowds or during
`
`assassination attempts or other crises where panic-struck crowds are particularly
`
`noisy or firefighters dispatched to sites of roaring conflagrations, or rescuers
`
`beneath a hovering helicopter--it is particularly important that the systems be
`
`5
`
`designed to adequately reject unwanted noise and maintain the requisite
`
`functionality permitting clear communication between talker and listener.
`
`One obvious approach to the suppression of ambient noise pickup has
`
`been to use a near-field microphone, such as a boom microphone, disposed in
`
`close proximity to the talker's mouth. With an appropriate mechanical or
`
`10
`
`electronic microphone configuration, sounds that originate at a distance to the
`
`microphone can be de-emphasized, such that only sounds originating in the
`
`vicinity of the microphone--i.e., those from the talker's mouth--are picked up for
`
`transmission. However, there are disadvantages to such configurations,
`
`including the need to place the microphone close to the mouth, thereby detracting
`
`15
`
`from the discreetness and convenience of the device while increasing its size and
`
`cost and the number of components used. Additionally, the user of the
`
`microphone must rotate the microphone into place before speaking and must be
`
`careful not to knock the headset supporting the microphone out of position or off
`
`his head during wear. All these problems detract from the utility of the device.
`
`20
`
`Another known type of sound pickup device relies on bone conduction. It
`
`is recognized that a talker's voice reaches the talker's own ear not only through
`
`the normal pathway of air pressure waves, but also through conduction by the
`
`jaw and other bones of the skull as induced by vibrations from the talker's sound(cid:173)
`
`generating apparatus during speech. Bone conduction involves conveyance of
`
`25
`
`speech-induced vibrations directly to the ear canal, or, for that matter, to any part
`
`- 2 -
`
`

`

`WO 00/21194
`
`PCT/US99/23234
`
`-3-
`
`of a person's body supported by bone structure. Since the bones provide a direct
`
`pathway from the sound source--that is, from the larynx and vocal chords for
`
`example--a device which is disposed against suitable bone structures of the talker
`
`so as to detect the bone vibrations is able to circumvent a noisy acoustic
`
`5
`
`environment.
`
`Another advantage to bone conduction derives from a talker's natural
`
`reaction to high ambient noise levels. Under normal circumstances, bone
`
`conduction is a poor medium for high frequency, low energy sound transmission,
`
`but is an excellent medium for low frequency, high energy transmission. Raising
`
`10
`
`the level of speech, by louder speaking in reaction to a noisy environment,
`
`imparts greater energy to the bones--especially information-bearing consonant
`
`energy which normally lies at the higher frequencies and to which, all things
`
`being equal, air pressure conduction is normally better suited than bone
`
`conduction. The talker, by raising his or her voice, is disproportionately raising
`
`15
`
`the sound level of the consonant component of speech, to which is attributed 80-
`
`90% of intelligibility, and is thus naturally compensating for the normally
`
`consonant-unfriendly nature of bone conduction by imparting greater energy
`
`precisely at the sound frequencies at which bone conduction is poor.
`
`The prior art has used vibration transducers and even accelerometers to
`
`20
`
`detect bone vibration. To that end, throat microphones were first introduced in
`
`WWII and worn by bomber pilots who needed to communicate in extremely
`
`noisy environments. The throat microphones picked up vibrations at the talker's
`
`throat and provided a clearer representation of his voice than a sound pressure
`
`microphone, which would have picked up the overshadowing ambient acoustic
`
`25
`
`noise as well. Such devices were crude and the quality of their signal poor.
`
`- 3 -
`
`

`

`WO 00/21194
`
`PCT /US99/23234
`
`-4-
`
`A device known as the Voiceducer™, manufactured by Tempco™, also
`
`takes advantage of bone conduction. It is placed in the user's ear canal and
`
`operates to pick up vibrations induced therein by the user's speech apparatus as
`
`described above. Similarly, Andrea Electronics™ manufactures a device having
`
`5
`
`two conventional air pressure microphones, each oriented differently from the
`
`other, and one of which is designed to pick up air pressure vibrations in the ear
`
`canal whose source is in fact due to bone vibration. The dual conversion relied
`
`upon by this device produces a further degradation of performance. Other prior
`
`art bone conduction pickup apparatus include bone vibration sensors disposed
`
`10
`
`directly atop the user's head, inertial microphones, and ceramic microphones
`
`operating without diaphragms.
`
`SUMMARY OF THE INVENTION
`
`The present invention exploits the phenomenon of bone conduction to
`
`suppress the deleterious effects of high acoustic noise environments without
`
`15
`
`suffering the degradation of voice pickup quality in low noise environments
`
`demonstrated by prior art bone conduction pickup systems. The above-described
`
`human speech characteristics are exploited by providing a device which
`
`adaptively senses sound through bone conduction when acoustic noise levels are
`
`high. As discussed above, bone conduction alone is generally a poor expedient
`
`20
`
`for transfer of intelligible speech, and in an acoustically quiet environment an air
`
`pressure microphone is much better suited to picking up a talker's voice. It is
`
`only when acoustic noise levels are high that the quality of air pressure
`
`microphone detection is degraded enough to make bone conduction a better
`
`alternative. The device in accordance with the invention accordingly relies on
`
`25
`
`both bone conduction and air pressure vibration to pick up the talker's voice.
`
`- 4 -
`
`

`

`WO 00/21194
`
`PCT /US99/23234
`
`-5-
`
`The bone conducted speech is detected using a motion transducer comprising a
`
`vibration sensor, and specifically, in a preferred embodiment, an accelerometer,
`
`although use of other types of vibration sensors, such as displacement and
`
`velocity sensors, is also contemplated. When ambient noise levels are low,
`
`5
`
`sound detection is effected primarily through a conventional microphone, which
`
`may comprise, inter alia, a directional microphone, a pressure difference
`
`microphone or an omni-directional microphone. An adjustable transition circuit
`
`is provided for fading out the conventional microphone signal and fading in the
`
`bone vibration sensor signal with an increase in ambient noise levels.
`
`10
`
`The transition circuit, in one embodiment, combines the signals from the
`
`microphone and vibration sensor, sensing the average amplitude of the
`
`microphone signal and inversely adjusting the gain of pre-amplifiers and/or
`
`amplifiers connected to the outputs of the vibration sensor and microphone. As
`
`the average amplitude of the microphone signal increases relative to a
`
`15
`
`predetermined threshold, due to increased ambient noise levels, the gain of the
`
`vibration sensor is increased while that of the microphone is decreased.
`
`Alternatively, the gain adjustment can be controlled by sensing the overall noise
`
`level detected by the vibration sensor rather than the microphone. Gain
`
`adjustment is effected such that the amplitude of the overall output signal is
`
`20
`
`maintained at a smoothly changing level throughout the transition range.
`
`In a second, preferred embodiment, the signal from the microphone is
`
`amplified linearly up to a predetermined level only, after which the signal level is
`
`limited and prevented from further increase. The signal from the vibration
`
`sensor, initially amplified at a lower gain level than that of the microphone, is not
`
`25
`
`limited and eventually, after the predetermined level, surpasses the limited
`
`- 5 -
`
`

`

`WO 00/21194
`
`PCT/US99/23234
`
`-6-
`
`microphone signal and comes to dominate, thereby improving system
`
`intelligibility and increasing immunity to ambient noise effects. Such a
`
`configuration can be implemented using available sound processing circuitry and
`
`off-the-shelf technology, at reduced production costs and power consumption
`
`5
`
`requirements.
`
`An additional advantage of the invention is its particular suitability to
`
`voice-activated applications, as the bone conduction expedient improves system
`
`immunity to false triggering due to noisy environments. The invention realizes
`
`power conservation advantages by for example turning off a radio transmitter in
`
`10
`
`the absence of the user's voice.
`
`Yet another advantage of the invention derives from the use of the two
`
`modalities of sound transfer--namely, air pressure waves and bone conduction.
`
`The two modalities permit the use of signal correlation techniques to enhance the
`
`voice signal, thus improving overall system performance. Characteristics of the
`
`15
`
`loudness of speech derived from signals from either or both modalities can be
`
`used to control the system of the invention. For example the amplitude envelope
`
`of the bone conducted signal can be used to dynamically vary the gain of the
`
`microphone amplifier, so that speech portions of the microphone signal are
`
`enhanced while noise portions are suppressed. These characteristics include
`
`20
`
`signal RMS value, peak value, average value, and energy level.
`
`The invention finds utility in numerous voice pickup applications,
`
`including but not limited to ear level pickups, telephony and dual microphone
`
`directional pickup systems.
`
`- 6 -
`
`

`

`WO 00/21194
`
`PCT /US99/23234
`
`-7-
`
`BRIEF DESCRIPTION OF THE DRAWINGS
`
`Many advantages of the present invention will be apparent to those skilled
`
`in the art with a reading of this specification in conjunction with the attached
`
`drawings, wherein like reference numerals are applied to like elements and
`
`5
`
`wherein:
`
`FIG. 1 is a schematic diagram of a dual-sensor device in accordance with
`
`the invention;
`
`FIG. 2 is a graphical illustration of the signal output of the transducers in
`
`one embodiment of the device of FIG. 1;
`
`10
`
`FIG. 3 is a schematic diagram of a preferred embodiment of a dual-sensor
`
`device in accordance with the invention;
`
`FIG. 4 is a graphical illustration of the signal output of the transducers in
`
`another embodiment of the device of FIG. 3;
`
`FIG. 5 is a schematic illustration showing an embodiment in accordance
`
`15
`
`with the invention in which selection between the accelerometer and microphone
`
`outputs is based on the detected envelopes of the signals;
`
`FIG. 6 is a schematic illustration in which phase correlation of the
`
`microphone and accelerometer signals is effected in accordance with the
`
`invention;
`
`- 7 -
`
`

`

`WO 00/21194
`
`PCT/US99/23234
`
`-8-
`
`FIGS. 7 and 8 are graphical illustrations of the use of signal compression
`
`and expansion contemplated by exemplary embodiments in accordance with the
`
`invention;
`
`FIG. 9 is a schematic illustration of an embodiment in accordance with
`
`5
`
`the invention in which the detected signals are processed by frequency band
`
`components;
`
`FIG. 10 is a schematic illustration of an embodiment in accordance with
`
`the invention in which variable filtering is used; and
`
`FIG. 11 is a graphical representation of the operation of the circuit of
`
`10
`
`FIG. 10.
`
`DESCRIPTION OF THE PREFERRED EMBODIMENTS
`
`FIG. 1 shows a schematic implementation of a dual-sensor device 10 in
`
`accordance with the invention. The device 10 uses an acoustic sound transducer,
`
`such as a microphone 20, and an accelerometer 30 whose outputs are selectively
`
`15
`
`manipulated by a processing circuit 16 to compensate for obfuscation of the
`
`talker's voice due to the presence of ambient acoustic noise. The device is held
`
`against the skin of the talker's skull and may take the form of an ear level pickup
`
`worn behind the ear, or a telephone handset held against the ear during
`
`telephonic communication. Other possibilities include placement in a hat band or
`
`20
`
`a face mask worn with, for example, a firefighter's helmet.
`
`- 8 -
`
`

`

`WO 00/21194
`
`PCT/US99/23234
`
`-9-
`
`Microphone 20 may be any conventional acoustic sound transducer such
`
`as, e.g., a standard microphone, directional microphone, omni-directional
`
`microphone, or pressure difference/gradient microphone or other microphone.
`
`Accelerometer 30 is a transducer device which senses the vibratory displacement
`
`5
`
`of the bones in proximity to the ear as imparted primarily by vibrations during
`
`speech. The output of accelerometer 30 is a second derivative representation of
`
`bone displacement exhibiting, in the frequency domain, a 12 dB/octave rise
`
`which matches the approximately 12 dB/octave roll-off of bone conduction
`
`speech signals at the ear. It is contemplated that vibration sensors other than
`
`10
`
`accelerometers may be used, and accordingly, use of the latter is not intended to
`
`be limiting. For instance, a velocity sensor exhibiting a first derivative rise in
`
`the frequency domain of 6 dB/octave can be used preferably with an electronic 6
`
`dB/octave frequency equalization, or similarly, a displacement sensor, may each
`
`be used in lieu of accelerometer 30 and also fall within the purview of the
`
`15
`
`invention.
`
`Signals from the transducers 20 and 30 are fed through corresponding
`
`equalization circuits 22 and 32, which operate to adjust the frequency response of
`
`the transducer signals and provide a flatter frequency response. Equalization
`
`circuits 22, 32 can be designed to provide greater gain at higher frequencies than
`
`20
`
`gain at lower frequencies to thereby decrease the "basiness" of the response,
`
`especially of accelerometer 30. It is also known that the bony structure of the
`
`head imparts certain resonances to the signal, so that the equalization is also
`
`intended to correct for those changes in the signal as well.
`
`Variable or adjustable gain pre-amplifiers 24 and 34 are connected to the
`
`25
`
`equalization circuits 22, 32. These pre-amplifiers, along with faders or
`
`- 9 -
`
`

`

`WO 00/21194
`
`PCT/US99/23234
`
`-10-
`
`amplifiers 26 and 36, serve to control the gain of the signals representing the
`
`transducer responses. An adder circuit 38 combines the signals from microphone
`
`20 and accelerometer 30 and provides the output signal 40 as the output signal of
`
`the pickup device through variable gain amplifier 46.
`
`5
`
`In operation, the output from the pre-amplifier 24 is applied to a control
`
`circuit 44 comprising for example an amplitude detector which senses the average
`
`amplitude of the signal from the microphone 20. When this average amplitude
`
`reaches or exceeds a predetermined threshold with a rise in ambient noise levels,
`
`the control circuit 44 commences a fading effect, decreasing the gain of amplifier
`
`10
`
`26 while increasing that of amplifier 36. In this manner, the bone conduction
`
`component of the audio signal, comprising primarily the talker's speech, is
`
`amplified to a greater extent than the ambient noise signal at high ambient noise
`
`levels. At low ambient noise levels, the microphone signal is permitted to
`
`dominate as the output signal. The microphone singal under lower ambient noise
`
`15
`
`levels provides a qualitatively better signal despite a measure of equalization
`
`applied to the accelerometer signal by equalization circuit 32 to improve the
`
`quality thereof. Typically, the value of the threshold will depend on the
`
`particular application and may vary anywhere in the range of and beyond 85 -
`
`100 dB SPL (sound pressure level--at 110 dB SPL ambient noise levels, a person
`
`20
`
`is unable to hear his or her own voice), while the transition region is selected to
`
`span around 20 dB, although this value is just exemplary, with the actual value
`
`being selected based on the particular application. Beyond the cross-over
`
`threshold, the signal from the accelerometer 30 will predominate, exceeding the
`
`microphone signal by about 20-30 dB, while preceding the threshold, the
`
`25
`
`microphone signal predominates. A graphical representation of the system signal
`
`output is depicted in FIG. 2, which shows the microphone signal dominating the
`
`- 10 -
`
`

`

`WO 00/21194
`
`PCT/US99/23234
`
`-11-
`
`accelerometer signal below a threshold level (shown by the dashed line), while
`
`the accelerometer signal dominates above the threshold. The transition region is
`
`the region where the transition between these two states occurs. Peak detector 42
`
`is provided to maintain the level of the overall signal constant, operating to adjust
`
`5
`
`the gains of pre-amplifiers 24 and 34 in accordance with the output 40.
`
`In the preferred embodiment in accordance with FIG. 3, a dual-sensor
`
`device 50 is shown in which gain control is effected only on the output signal of
`
`one of the transducers--the microphone 70. Processing circuit 56 uses existing
`
`sound processing circuitry, depicted within the dashed region 60, in conjunction
`
`10
`
`with accelerometer 80, equalization circuit 82, pre-amplifier 84, and adder 76 to
`
`provide the enhanced voice pickup device of the invention. Circuit 60 comprises
`
`equalization circuit 72, AGC (automatic gain control) amplifier circuit 74 having
`
`a specific control circuit (not shown) for controlling the gain thereof, and a
`
`bandsplit filter device 78 per known multi-band full dynamic range compression
`
`15
`
`sound processing technology used in, e.g., hearing aid technology and ideally
`
`suited for operation at low voltage and current requirements.
`
`The dual-sensor 50 operates as follows. The output of microphone 70 is
`
`provided to equalization circuit 72 which again serves to flatten the frequency
`
`response for improved quality. The signal is then fed to AGC circuit 74, which
`
`20
`
`is designed to have a gain limiting threshold at around 85 dB SPL. Circuit 74
`
`operates as a linear amplifier, limiting amplification beyond the selected
`
`threshold, which in this implementation is 85 dB SPL such that the output signal
`
`simply will not become louder with an increase of the input signal. It is to be
`
`noted that limiting is different from clipping in that clipping, which eliminates the
`
`25
`
`peaks of the signals above a predetermined threshold, consequently introduces
`
`- 11 -
`
`

`

`WO 00/21194
`
`PCT /US99/23234
`
`-12-
`
`undesirable distortions of the sound signal. Signal limiters are well known in the
`
`art.
`
`By contrast, the gain of pre-amplifier circuit 84 is not limited, and as
`
`ambient noise levels increase, the accelerometer signal, unlike the microphone
`
`5
`
`signal, continues to increase. Although initially set at a lower level than that of
`
`the microphone signal, the accelerometer signal eventually surpasses that of the
`
`microphone when the ambient noise level reaches a sufficient magnitude because
`
`the talker naturally raises his voice in order to hear himself. Adder 76 operates
`
`to combine the signals, with output from the microphone predominating at the
`
`10
`
`low ambient noise level and the output from the accelerometer predominating at
`
`the high ambient noise level. Again, the accelerometer is exemplary and can be
`
`any vibration sensor.
`
`The characteristic signal output is graphically represented in FIG. 4. As
`
`in the previous embodiment, the microphone signal output dominates at sound
`
`15
`
`levels below the cross-over threshold, while the accelerometer signal dominates
`
`at levels above the transition threshold.
`
`The dual-sensor 50 is designed with readily available circuitry and off(cid:173)
`
`the-shelf components. It has minimal voltage and current consumption
`
`characteristics and is well-suited for miniaturized, in-the-ear applications as
`
`20
`
`contemplated for ear level pickup devices. Inherent advantages afforded thereby
`
`include reduced cost and simplicity of design and assembly, as well as enhanced
`
`performance provided by the unique application of the two transducers. Also, as
`
`noise increases, the ratio of accelerometer signal to microphone signal improves
`
`- 12 -
`
`

`

`WO 00/21194
`
`PCT /US99/23234
`
`13-
`
`i.e., noise output becomes progressively a smaller and smaller portion of output
`
`giving a progressively better signal-to-noise ratio.
`
`FIG. 5 shows an embodiment in accordance with the invention in which
`
`5
`
`selection between the accelerometer and microphone outputs is based on the
`
`detected envelopes of the signals. Signals from microphone 92 are provided to
`
`pre-amplifier 93 and envelope detector 94, while signals from accelerometer 96
`
`are provided to pre-amplifier 97 and envelope detector 98. A switching circuit
`
`95 selects the more variable of these signals.
`
`It should be noted that while the
`
`10
`
`noise envelope is nearly constant, the voice envelope is greatly time-varying as
`
`noise increases. Consequently, the microphone signal will vary less but the
`
`accelerometer signal will continue to vary. In lower noise, microphone signal
`
`will vary more than accelerometer signal due to low-pass nature of bone
`
`conduction. The envelope detectors 94 and 98 can have short (-10 ms) time
`
`15
`
`constants for selection of individual syllables, or long (-1 s) so gradual selection
`
`is made as background noise varies. Additionally, the switching circuit 95 can
`
`be replaced preferably with a fader circuit to permit gradual rather than abrupt
`
`transitions.
`
`FIG. 6 alternatively shows phase correlation of the microphone and
`
`20
`
`accelerometer signals in accordance with the invention. Signals from the
`
`microphone 100 and accelerometer 130 are applied to correlator 120 via pre(cid:173)
`
`amplifiers 110 and 140, respectively, after the phase difference of the
`
`accelerometer signal is corrected in phase difference correction circuit 150. By
`
`thus correlating the phase corrected signals, either on an average energy basis,
`
`25
`
`instantaneous signal level basis or in individual multi-frequency bands (via Fast
`
`Fourier Transformation implemented by digital signal processing, for example),
`
`- 13 -
`
`

`

`WO 00/21194
`
`PCT/US99/23234
`
`-14-
`
`the correlated components of the signals can be passed to the output, whereas the
`
`uncorrelated components can be determined to be noise and prevented from
`
`passing to the output. In other words, the determination can be made either on a
`
`time-basis only or on both a frequency and time-basis simultaneously.
`
`5
`
`In an alternative embodiment, AGC amplifier 74 and pre-amp 84 of FIG.
`
`3 can be configured to provide non-linear amplification, controlled by an
`
`appropriate control circuit, such that AGC amplifier 74 provides signal
`
`compression at a predetermined ratio, while pre-amp 84 provides signal
`
`expansion at a predetermined ratio. Various combinations of amplification
`
`10
`
`characteristics relative to a transition threshold can be used. More particularly,
`
`as shown in FIG. 7, using the AGC amplifier 74, the microphone signal can be
`
`amplified at a linear, 1: 1 ratio below the transition threshold and compressed at a
`
`N:l ratio (N> 1, e.g., 2.5, 10, etc.) above the transition threshold, while the
`
`accelerometer signal, using pre-amplifier 84, is expanded at a 1 :M ratio (M > 1,
`
`15
`
`e.g., 2 or 1.5, etc.) throughout the operational range. Alternatively, as shown in
`
`FIG. 8, the microphone signal can be amplified at a linear 1:1 ratio below the
`transition threshold and compressed at a N: 1 (N > 1) ratio above the transition
`threshold, while the accelerometer signal is expanded at a 1 :M ratio (M > 1)
`
`below the transition threshold and linearly amplified at a 1: 1 ratio above the
`
`20
`
`threshold. The amplifiers are designed to produce equal output levels at the
`
`transition SPL level. Those skilled in the art will recognize that many different
`
`combinations of linear amplification, compression and expansion of the detected
`
`signals are possible and would depend on the particular application. Moreover,
`
`the transition decision, made by the control circuit, can be linked to any
`
`25
`
`characteristic of one or both detected signals, including RMS values, peak
`
`values, average values, signal energy, correlation and variability, again
`
`- 14 -
`
`

`

`WO 00/21194
`
`PCT/US99/23234
`
`15
`
`depending on the particular application contemplated and the physical and
`
`economic constraints imposed.
`
`In another embodiment in accordance with the invention and
`
`schematically shown in FIG. 9, the signals from the microphone and vibration
`
`5
`
`sensor can be split into high and low frequency bands and then faded separately.
`
`In FIG. 9, the signal from microphone 160 is amplified by amplifier 162 and then
`
`applied to bandsplit filter 164 serving to split the signal into a high frequency
`
`band and a low frequency band. Similarly, the signal from vibration sensor 166
`
`is amplified by amplifier 168 and then applied to bandsplit filter 170 serving to
`
`10
`
`split the signal into high and low frequency bands. Bandsplit filters 164 and 170
`preferably have the same critical frequency t The signals are then respectively
`applied to high frequency and low frequency fader circuits 172 and 174, the
`
`outputs of which are summed in adder 176 and provided as the system output.
`
`Fader circuits 172 and 174 comprise any circuit discussed above designed to
`
`15
`
`implement the fade over functions between the microphone and the vibration
`
`sensor.
`
`The circuit of FIG. 9 is well-suited to exploit the recognized phenomenon
`
`that most background acoustic noise--the drone of machinery, the roar of a
`
`blazing fire, etc.--lies at the low frequency end of the noise spectrum. By
`
`20
`
`splitting the input signal into low and high frequency band components, fading is
`
`effected earlier--i.e., at a lower overall noise intensity level--for the low
`
`frequency band component than for the high frequency band component simply
`
`because the high frequency band contains less energy than the low frequency
`
`band. This maintains more of the high frequency intelligibility which passes
`
`25
`
`through the acoustic path to the microphone 160 while still removing much of the
`
`- 15 -
`
`

`

`WO 00/21194
`
`PCT/US99/23234
`
`16
`
`excess bass noise due to the earlier fade over in the low frequency band to the
`
`cleaner bone conduction signal of vibration sensor 166, which preferably
`
`comprises an accelerometer.
`
`The detected sound spectrum can be split into more than two bands, each
`
`5
`
`of which is faded at a corresponding intensity level. Digital signal processing
`
`techniques readily lend themselves to such an implementation, with Fast Fourier
`
`Transform techniques being used in conjunction with suitable algorithms for
`
`effecting the fading when the signal intensity at a band exceeds a threshold
`
`intensity. The threshold intensity would be determined at the level at which the
`
`10
`
`microphone 160 signal indicates there is so much noise that the voicing is no
`
`longer passing through, and the signal in that band would then be faded over to
`
`bone conduction from vibration sensor 166. The threshold level would be
`
`predetermined and device-specific, depending on the characteristics of the overall
`
`system design and the particular application for the device, but would generally
`
`15
`
`be set at about 95 -100 dB SPL noise level.
`
`A particular advantage of such a multi-band processing system is realized
`
`when the interfering noise is at specific regions in the nois