• IEEE
`
`Advancing Technology
`for Humanity
`
`DECLARATION OF GERARD P. GRENIER
`
`I, Gerard P. Grenier, am over twenty-one (21) years of age. I have never been convicted
`of a felony, and I am fully competent to make this declaration. I declare the following to be true
`to the best of my knowledge, information and belief:
`
`1. I am Senior Director of Content Management of The Institute of Electrical and
`Electronics Engineers, Incorporated ("IEEE").
`
`2. IEEE is a neutral third party in this dispute.
`
`3. Neither I nor IEEE itself is being compensated for this declaration.
`
`4. Among my responsibilities as Senior Director of Content Management, I act as a
`custodian of certain records for IEEE.
`
`5. I make this declaration based on my personal knowledge and information contained
`in the business records of IEEE.
`
`6. As part of its ordinary course of business, IEEE publishes and makes available
`technical articles and standards. . These publications are made available for public
`download through the IEEE digital library, IEEE Xplore.
`
`7. It is the regular practice of IEEE to publish articles and other writings including
`article abstracts and make them available to the public through IEEE Xplore. IEEE
`maintains copies of the publications in the ordinary course of its regularly conducted
`activities.
`
`8. The article below has been attached as Exhibit A to this declaration:
`
`A. Qi Li, et al., "A portable USB-based microphone array device for robust
`speech recognition", 2009 IEEE International Conference on Acoustics,
`Speech and Signal Processing, April 19 -24, 2009.
`
`9. I obtained a copy of Exhibit A through IEEE Xplore, where it is maintained in the
`ordinary course of IEEE's business. Exhibit A is a true and correct copy of the
`Exhibit, as it existed on or about April 2, 2020.
`
`IEEE
`1 O. The article and abstract from IEEE Xplore shows the date of publication.
`Xplore populates this information using the metadata associated with the publication.
`Amazon Ex. 1008
`IPR Petition – US RE47,049
`
`445 Hoes Lane Piscataway, NJ 08854
`
`Amazon Ex. 1008, Page 1 of 9
`
`

`

`11. Qi Li, ct al., "A p rt b1
`d microphone arr y devic for robust peech
`recognition" was publish din th 200
`lnternadonal on erenc on Acoustics,
`I
`Speech and
`ignal Proc
`ing proc ding .
`e 2009 I
`International Conference
`on Acoustics, Sp cch and
`ignal
`roe ssing was h ld rom April 19 - 24, 2009.
`Copies of the conferenc proc dings w re mad available no later than th last day
`of the conferenc . The articl
`is currently availabl for public download from the
`IEEE digital library, I
`Xplorc.
`
`12. I hereby declare that all statements made herein of my own knowledge are true and
`that all statements made on information and belie are believed to be true, and further
`that these statements were made with the knowledge that willful false statements and
`the like are punishable by fine or imprisonment, or both, under 18 U.S.C. § 1001.
`
`I declare under penalty of perjury that the foregoing stalemell~:t:: :a~corr:
`&_ ---=----
`Executedon: ~-/-Jpr,/-- dD;},Q
`
`•
`
`Amazon Ex. 1008, Page 2 of 9
`
`

`

`
`
`
`
`
`
`EXHIBIT A
`
`EXHIBIT A
`
`Amazon EX. 1008, Page 3 0f 9
`
`Amazon Ex. 1008, Page 3 of 9
`
`

`

`
` IEEE.org IEEE Xplore Digital Library
`|
`
`
`|
`
`IEEE-SA
`
`
`|
`
`IEEE Spectrum
`
`
`|
`
`More Sites
`
`
` Cart Create Account
`|
`
`
`|
`
` Personal Sign In
`
`Institutional Sign In
`
`Browse
`
`My Settings
`
`Get Help
`
`Subscribe
`
`Conferences > 2009 IEEE International Confe...
`
`A portable USB-based microphone array device for robust
`speech recognition
`Publisher: IEEE
`
`Cite This
`
`
`
`Cite This
`
` PDF
`
`Alerts
`
`Manage
`Content Alerts
`
`Add to
`Citation Alerts
`
`3 Author(s)
`
`Qi Li ; Manli Zhu ; Wei Li View All Authors
`
`197
`Full
`Text Views
`
`1P
`
`atent
`Citation
`
`3P
`
`aper
`Citations
`
`Abstract
`
`Document Sections
`
`1.
`
`Introduction
`
`2. System
`Description
`
`2. Broadband
`Beamforming
`
`3. Noise Reduction
`
`5. ASR Experimental
`Results
`
`Authors
`
`Figures
`
`References
`
`Citations
`
`Keywords
`
`Metrics
`
`More Like This
`
`Abstract: We present a USB-based, highly directional, and portable microphone array
`device that delivers a crisp, clear and noise-reduced speech signal. This device consists
`of fou... View more
`
` Metadata
`Abstract:
`We present a USB-based, highly directional, and portable microphone array device that
`delivers a crisp, clear and noise-reduced speech signal. This device consists of four
`linearly distributed microphone sensors and a filter-and-sum beamformer designed
`using broadband beam-forming algorithm. The device has a narrow acoustic beam
`pattern and identical frequency responses for almost all speech bands. In addition to
`beamforming, an adaptive noise reduction algorithm is used to further reduce the
`background noise. By utilizing both the spatial and temporal information, the SNR of
`speech signals is improved and speech recognition performance in noisy environments
`is significantly improved as reported in our experiments.
`
`Published in: 2009 IEEE International Conference on Acoustics, Speech and Signal
`Processing
`
`Date of Conference: 19-24 April 2009
`
`INSPEC Accession Number: 10701439
`
`Date Added to IEEE Xplore: 26 May 2009
`
`DOI: 10.1109/ICASSP.2009.4959830
`
` ISBN Information:
`
` ISSN Information:
`
`Publisher: IEEE
`
`Conference Location: Taipei, Taiwan
`
` Export to
`
`Collabratec
`
`Downl
`PDF
`
`More Like This
`
`Microphone array signal processing for far-
`talking speech recognition
`2001 IEEE Third Workshop on Signal
`Processing Advances in Wireless
`Communications (SPAWC'01). Workshop
`Proceedings (Cat. No.01EX471)
`Published: 2001
`
`A microphone array system for speech
`recognition
`1997 IEEE International Conference on
`Acoustics, Speech, and Signal Processing
`Published: 1997
`
`View More
`
`Top Organizations with Patents
`on Technologies Mentioned in
`This Article
`
`
`
`Amazon Ex. 1008, Page 4 of 9
`
`

`

` Contents
`
`1. Introduction
`A microphone array consists of multiple microphone sensors located at
`different positions. It can be used for both sound source location [1] and
`speech enhancement by processing the signals from each individual
`sensors [2] [3]. While most of the current speech processing software
`Sign in to Continue Reading
`can only use the temporal information, the designed device utilizes both
`the spatial and temporal information; thus, significantly improving
`speech recognition performance and robustness.
`
`Authors
`
`Figures
`
`References
`
`Citations
`
`Keywords
`
`Metrics
`
`
`
`
`
`
`
`
`
`
`
`
`
`IEEE Personal Account
`
`Purchase Details
`
`Profile Information
`
`Need Help?
`
`CHANGE USERNAME/PASSWORD
`
`PAYMENT OPTIONS
`
`COMMUNICATIONS PREFERENCES
`
`US & CANADA: +1 800 678 4333
`
`VIEW PURCHASED DOCUMENTS
`
`PROFESSION AND EDUCATION
`
`WORLDWIDE: +1 732 981 0060
`
`TECHNICAL INTERESTS
`
`CONTACT & SUPPORT
`
`Follow
`
`  
`
`About IEEE Xplore | Contact Us | Help | Accessibility | Terms of Use | Nondiscrimination Policy | Sitemap | Privacy & Opting Out of Cookies
`A not-for-profit organization, IEEE is the world's largest technical professional organization dedicated to advancing technology for the benefit of humanity.
`
`© Copyright 2020 IEEE - All rights reserved. Use of this web site signifies your agreement to the terms and conditions.
`
`IEEE Account
`
` Purchase Details
`
` Profile Information
`
` Need Help?
`
`» Change Username/Password
`» Update Address
`
`» Payment Options
`» Order History
`» View Purchased Documents
`
`» Communications Preferences
`» Profession and Education
`» Technical Interests
`
`» US & Canada: +1 800 678 4333
`» Worldwide: +1 732 981 0060
`» Contact & Support
`
`
` About IEEE Xplore Contact Us
`|
`
`
`|
`
`Help
`
`
`|
`
`Accessibility
`
`
`|
`
`Terms of Use
`
`
`|
`
`Nondiscrimination Policy
`
`
`|
`
`Sitemap
`
`
`|
`
`Privacy & Opting Out of Cookies
`
`A not-for-profit organization, IEEE is the world's largest technical professional organization dedicated to advancing technology for the benefit of humanity.
`© Copyright 2020 IEEE - All rights reserved. Use of this web site signifies your agreement to the terms and conditions.
`
`Amazon Ex. 1008, Page 5 of 9
`
`

`

`978-1-4244-2354-5/09/$25.00 ©2009 IEEE
`
`1301
`
`ICASSP 2009
`
`A PORTABLE USB-BASED MIROPHONE ARRAY DEVICE FOR ROBUST SPEECH
`RECOGNITION
`
`Qi (Peter) Li, Manli Zhu, and Wei Li*
`
`Li Creative Technologies, Inc. (LcT)
`30 A Vreeland Road, Suite 130, Florham Park, NJ 07932, USA
`E-mail: li,manlizhu@licreativetech.com; *iewil2000@yahoo.com
`
`
`challenge of a microphone array compared to single close-
`talking microphones is the decreased performance in speech
`recognition because of the variation of their frequency
`responses [8]. These problems need to be addressed by
`improving the array design algorithms, in order to develop a
`portable and high performance device.
`
`
`
`
`
`
`
`
`Figure 1. The 4-sensor microphone array named CrispMicTM
`clipped on a laptop
`
`
`In this paper, we present a portable microphone array
`device, named CrispMic™, where both the size and
`performance have been properly addressed. As shown in
`Figure 1, the size of the microphone array is only 9.6 x 2.6 x
`1.3 cm which is less than the half of the size of a similar
`microphone array on the market. Also, it has a constant
`frequency response in a wide range of speech bands as
`shown in Figure 4. The experimental results in this paper
`show that the array can improve speech recognition
`performances
`significantly,
`especially
`in
`noisy
`environments.
`
`
`2. SYSTEM DESCRIPTION
`
`
`
`Microphone
`amplifier
`Microphone
`amplifier
`Microphone
`amplifier
`
`Microphone
`amplifier
`
`
`ADC
`
`Flash memory
`
` DSP
`Beam-
`Noise
`Reduction
`forming
`
`USB
`
`Figure 2. Illustration of the hardware structure
`
`Figure 2 shows the structure of the device. It consists of four
`identical omni-directional microphones, an audio codec chip
`with an ADC and pre-amplifiers, a DSP (digital signal
`
`
`
`ABSTRACT
`
`
`We present a USB-based, highly directional, and portable
`microphone array device that delivers a crisp, clear and
`noise-reduced speech signal. This device consists of four
`linearly distributed microphone sensors and a filter-and-sum
`beamformer designed using broadband beam-forming
`algorithm. The device has a narrow acoustic beam pattern
`and identical frequency responses for almost all speech
`bands. In addition to beamforming, an adaptive noise
`reduction algorithm is used to further reduce the background
`noise. By utilizing both
`the spatial and
`temporal
`information, the SNR of speech signals is improved and
`speech recognition performance in noisy environments is
`significantly improved as reported in our experiments.
`
`Index Terms — Microphone array, beamforming, noise
`reduction, robust speech recognition.
`
`
`1. INTRODUCTION
`
` A
`
` microphone array consists of multiple microphone
`sensors located at different positions. It can be used for both
`sound source location [1] and speech enhancement by
`processing the signals from each individual sensors [2][3].
`While most of the current speech processing software can
`only use the temporal information, the designed device
`utilizes both the spatial and temporal information; thus,
`significantly improving speech recognition performance and
`robustness.
`
`One of the major challenges in applying a microphone array
`in speech recognition is that speech is a wideband signal.
`The traditional narrowband beamforming techniques are not
`appropriate anymore [4]. The problem has been addressed
`by the spatial Fourier transform of a continuous aperture [5]
`and the joint optimization of the spatial and frequency
`response [6]. In these approaches, to keep the constant
`response over the wide frequency range, the array size is
`usually large; thus most of the prototypes or products using
`microphone arrays on the market are quite large and cannot
`be used as a portable device [7]. The large size of the array
`prevents the array products from broad applications, such as
`handheld devices, wireless handsets, and PDA. Another
`
`Amazon Ex. 1008, Page 6 of 9
`
`

`

`1302
`
`n(cid:166) (cid:32)
`W
`X
`(
`),
`(
`)
`(cid:84)(cid:90)(cid:90)
`(
`),
`(cid:84)(cid:90)
`n
`1
`X
`),
`),
`(
`(
`(cid:84)(cid:90)
`(cid:84)(cid:90)
`where X is the signal received at the center of the array and
`W is the frequency response of the real-valued FIR filter w.
`If the sound source is far enough from the array, the
`the nth
`difference between
`the signal
`received by
`microphone xn and the center of the array is a pure delay.
`df
`c
`
`
` / cos (cid:84)
`(cid:87) (cid:32)
`We use
` to measure the delay by the
`n
`s
`n
`number of samples, where fs is the sampling frequency, c is
` is the microphone
`sound speed, and
`X
`X
` e),(
`
`j
`),(
`
`
` (cid:90)(cid:87)(cid:84)(cid:90)(cid:87)(cid:90) (cid:16)(cid:32)
`n
`signal. The spatial directivity pattern H can be re-written as:
`
`(cid:32) (cid:166) (cid:32)
`
`4
`
`H
`W
`,(
`(
`)
`(cid:84)(cid:90)
`(cid:90)
`n
`n
`
`where wT = [w1
`vector.
`
`Let the desired spatial directivity pattern equal 1 in the pass
`band and 0 in the stop band. The cost function is then
`defined as:
`
`
`
`
`2
`
`p
`
`s
`
`(cid:179)
`(cid:179)
`(cid:179) (cid:179)
`wJ )(
`dd
`dθ
`
`H
`
`H
`d
`
`|
`,(
`|
`,(
`|)
`|1)
`2
`(cid:90)(cid:84)(cid:90)
`(cid:32)
`(cid:16)
`
` (cid:14)(cid:68)(cid:84)(cid:90)
`(cid:84)(cid:90)
`(cid:58) (cid:52)
`(cid:58) (cid:52)
`p
`s
` (3)
`
` / wJ
`Let
`. We can then obtain the best parameter set
`0
`(cid:119)
`(cid:32)(cid:119)
`w.
`
`A Computer simulation was conducted to verify the
`performance of our designed beamformer with the following
`parameters: The distance between microphones is 0.02m,
`the sampling frequency fs = 48k, and FIR filter taper length
`L=128. When the pass-band (Θp, Ωp) = {300-4000Hz, 70o-
`110o}, the designed spatial directivity pattern is 1. When the
`stop-band (Θs, Ωs) = {300~4000Hz, 0o~60o+120o~180o}, the
`designed spatial directivity pattern is 0.
`
` (1)
`
`4
`n
`
`(cid:32)
`
`XY
`
`
`
`H
`
`(
`),
`(cid:84)(cid:90)
`
`(cid:32)
`
`n
`
`(cid:16)
`
`j
`)(
`(cid:84)(cid:90)(cid:87)
`n
`
`)e
`
`(cid:32)
`
`gwT
`
`,(
`)
`(cid:84)(cid:90)
`
` (2)
`
`1
`
`T,w2
`
`T,w3
`
`T,w4
`
`T] and g(ω,θ) is the steering
`
`processor) chip, a flash memory and a USB interface. The
`acoustic signal is picked up by four microphone components
`arranged as a linear array with 20 mm intervals. The audio
`codec provides an adjustable gain, and converts the four
`channels of analog signals into digital signals for the DSP.
`The beamforming algorithm combines the 4 channels of
`speech into one channel and then a noise reduction
`algorithm is applied to further reduce the background noise.
`The processed clean speech signals are then transmitted to a
`laptop or PC through the USB interface.
`
`The flash memory stores the software code for the DSP
`chip. Once the system boots up, the DSP chip reads the code
`from the flash memory into the internal memory and starts
`to execute the code. The device is powered through the
`USB. It is a plug-and-play device and can be used without
`installing any software. Since both the analogue and digital
`circuits are implemented in a small PCB, the circuit board
`was especially designed to avoid the noise interferences.
`
`
`2. BROADBAND BEAMFORMING
`
`
`Due to the special requirements in size and performance, we
`developed a robust, far-field broadband beamforming
`algorithm and implemented it in the DSP chip. The
`beamformer has a constant response in the speech frequency
`bands between 300-4000Hz. In theory, it can significantly
`improve spatial SNR of
`the speech signals without
`distortions in different frequency bands.
`
`The linear microphone array configuration is shown in
`Figure 3. It is comprised of four equally spaced microphone
`sensors, where di is the distance between the ith microphone
`and the center of the array. The output y of the array is the
`the four microphone outputs, y =
`filter-and-sum of
`(cid:166) (cid:32)
`4
`T
`n xw n
`.
`1n
`
`
`
`Sound Source
`
`τ3
`0
`
`d2
`
`d3
`
`θ
`d4
`
`d1
`
`x1
`w1
`
`x2
`w2
`
`x3
`w3
`
`x4
`w4
`
` ∑
`
`y
`
`
`
`Frequency 300 Hz
`to 4 KHz
`
`
`
`
`Figure 3. The configuration of linear microphone array.
`
`The spatial directivity pattern H(ω,θ) for the sound source
`form angle θ with normalized frequency ω is defined as [2]:
`
`Directional angles: from
`0 to 180 degrees
`
`
`Figure 4. Directivity pattern of the designed microphone array: the
`frequency bands from 300 Hz to 4 kHz of the sound from the front
`of the microphone array are enhanced and the sound from other
`directions are reduced by about 15 dB.
`
`Amazon Ex. 1008, Page 7 of 9
`
`

`

`1303
`
`
`
`(a)
`
`(b)
`
`(c)
`
`(d)
`
`
`
`Figure 4. Computer simulation results for the designed 4-sensor
`microphone array: The directional response for frequencies of
`(a) 0.5, (b) 1.0, (c) 2.0, (d) 3.0, and (e) 4.0 KHz.
`
`
`(e)
`
`
` (a) (b) (c)
`
`
`
`(e)
`(d)
`
`Figure 5. The directivity pattern of the microphone array based on
`our measurement in an anechoic chamber: The direction response
`for frequencies of (a) 0.5, (b) 1.0, (c) 2.0, (d) 3.0 and (e) 4.0 KHz.
`(The center is 0 dB.) The directional gains are significant and
`match the theoretical design and simulation.
`Figure 4 shows
`the directivity pattern of designed
`microphone array. In all frequency bands, the main lobe has
`the same level, which means the speech signal has little
`distortion in frequency. The main lobe is about 15dB higher
`than the side lobe; therefore the background sound from
`other directions will be highly suppressed compared to the
`sound in the desired pass direction.
`
`the design simulation, we conducted an
`To verify
`to measure
`the spatial response of
`the
`experiment
`microphone array in an anechoic chamber using a white
`noise stimulus. The microphone array was placed in a fix
`positioned and rotated in 20 degree increments. For each
`rotation, we applied the designed filter wi, i = 1,2,3 and 4, on
`the signal picked by each microphone and calculated the
`energy gain. The spatial directivity pattern is shown in
`Figure 5. The main lobe has a similar width among all the
`bands. The result shows that the directional gains are
`significant and are similar to our theoretical simulation.
`
`
`
`3. NOISE REDUCTION
`
`Our adaptive noise reduction algorithm consists of three key
`components: frequency analysis, adaptive Wiener filtering,
`and frequency synthesis. The frequency-analysis component
`is used for transforming the wideband noisy speech
`sequence into the frequency domain so that the subsequent
`analysis can be performed on a sub-band basis. This is
`achieved by the short-time discrete Fourier transform
`(DFT). The output from each frequency bin of the DFT
`represents one new complex valued time-series sample for
`the sub-band frequency range corresponding to that bin. The
`band width of each sub-band is given by the ratio of the
`sampling frequency to the transformed length.
`
`The most crucial step of the temporal filtering technique is
`an adaptive Wiener filter, which estimates the clean-speech
`spectrum from the noisy-speech spectrum. The system
`explores the short-term and long-term statistics of noise and
`speech, as well as the segmental SNR, to support a Wiener
`gain filtering. The noisy-speech spectrum passes through the
`Wiener filter, which then generates an estimate of the clean-
`speech spectrum. In the last step, the frequency synthesis, as
`an inverse process of the frequency analysis, reconstructs
`the clean-speech signal given the estimated clean-speech
`spectrum.
`
`
`5. ASR EXPERIMENTAL RESULTS
`
`
`In order to verify the performance of the CrispMic™, we
`have conducted a
`sequence of experiments. The
`experimental configuration is shown in Figure 6. The
`recorded speech was played by an artificial mouth while the
`background noise was played by four loudspeakers.
`
`Speaker 2
`6’; 45o
`
`Speaker 3
`5’; 50o
`
`Speaker 1
`
`Artificial Mouth
`
`Speaker 4
`
`5’6”; 30o
`
`15”
`
`4’20”; 20o
`
`
`
`Figure 6. Experimental configuration
`
`Experiment 1: Figure 7 shows the SNR improvement by
`using the CrispMic™. In this specific experiment, speech
`was played by an artificial mouth while white noise was
`played from Speaker 4 only as shown in Figure 6. Figure 7
`(a) is the signal recorded by traditional omni-directional
`microphone, where the SNR is 3.3dB. Figure 7 (b) is the
`signal processed by beamforming. The SNR is increased
`about 31dB by the spatial approach. Figure 7 (c) is the
`
`Amazon Ex. 1008, Page 8 of 9
`
`

`

`1304
`
`signal after noise reduction, which is also the final output of
`the CrispMic™. SNR is further improved another 10dB by
`the temporal approach.
`
`
`0.1
`
`0.08
`
`0.06
`
`0.04
`
`0.02
`
`0
`
`-0.02
`
`-0.04
`
`-0.06
`
`-0.08
`
`0
`
`
`
`0.5
`
`1
`
`1.5
`
`2
`
`2.5
`
`3
`
`3.5
`
`4
`
`4.5
`
`5
`x 105
`
`(a)
`
`0.1
`
`0.08
`
`0.06
`
`0.04
`
`0.02
`
`0
`
`-0.02
`
`-0.04
`
`-0.06
`
`-0.08
`
`-0.1
`
`0
`
`0.5
`
`1
`
`1.5
`
`2
`
`2.5
`
`3
`
`3.5
`
`4
`
`4.5
`
`5
`x 105
`
`
`
`0.1
`
`0.08
`
`0.06
`
`0.04
`
`0.02
`
`0
`
`-0.02
`
`-0.04
`
`-0.06
`
`-0.08
`
`-0.1
`
`0
`
`0.5
`
`1
`
`1.5
`
`2
`
`2.5
`
`3
`
`3.5
`
`4
`
`4.5
`
`5
`x 105
`
`
`
` (c)
` (b)
`Figure 7. The improvements on SNR: (a) The original signal
`captured by a traditional mic, SNR = 3.3; (b) The signal after
`beamforming, SNR=36.9dB; and (d) The signal after noise
`reduction, SNR = 49.4.
`
`
`CrispMic™
`
`CrispMicTM w/o
`noise reduction
`
`Traditional Mic
`
`Comparison of ASR
`Performance
`
`
`
`Figure 8. Comparison of ASR performance
`
`
`Experiment 2: The CrispMic™ has also undergone a series
`of tests in the automatic speech recognition system and the
`performance is measured in terms of the recognition
`accuracy. In our experiments, background helicopter noise
`was played by four loudspeakers, Speakers 1-4 as in Figure
`6. The CrispMic™ was connected to a laptop where
`Nuance’s
`software was used
`for
`recognition. The
`experiments used 244 pre-recorded English phrases. During
`the test, if any word in the recognized phrase was different
`from the spoken phrase, we counted it as an error and the
`entire phrase was rejected. The experiment allowed for two
`attempts; if the first attempt failed, we allowed a second
`attempt using the same phrases to simulate the real
`applications. Our experimental results are plotted in Figure
`
`8 and Table 1. The results showed that the contributions of
`our microphone array and noise reduction to speech
`recognition are very noteworthy.
`
`
`Table 1. Comparison of Speech Recognition Performances on the
`CrispMic™: The number of loudspeakers is 4, the number of
`attempts is 2, and the number of test phrases is 244.
`SNR (dB)
`0.40
`2.66 7.44 11.99 15.01
`Traditional microphone
`77.0
`77.9
`82.7
`87.2
`93.4
`(%)
`4-microphone
`LcT:
`array only (%)
`CrispMic™: 4-mic array
`plus noise reduction (%)
`
`
`84.6
`
`85.6
`
`93.0
`
`95.4
`
`91.5
`
`92.3
`
`94.2
`
`95.5
`
`97.5
`
`98.7
`
`4. CONCLUSIONS
`
`In this paper, we presented a novel microphone array
`device. Compared to traditional microphones, the device
`utilizes both the spatial and temporal information; therefore
`it significantly improved SNR and speech recognition
`performance. Compared to the similar product or prototype
`on the market, this device is much smaller in size, which
`facilitates its application to portable devices, such as
`wireless phones, PDA, and Laptops.
`
`
`5. ACKNOWLEDGEMENT
`
`
`thank Jingdong Chen for useful
`to
`like
`We would
`discussions and Ashrith Deshpanda, Romnel Crasta, and
`Xiaoqian Xiao for conducting some of the experiments.
`
`
`6. REFERENCES
`
`[1] C.H. Knapp and G.C. Carter, The Generalized Correlation
`Method for Estimation of Time Delay, IEEE Trans. on
`Acoustic, Speech and Signal Processing, Vol.ASSP-24, No. 4,
`pp.320-327, Aug.1976.
`[2] S. Doclo, M. Moonen, Design of Far-Field and Near-Field
`Signal
`Broadband Beamformers Using Eigenfilters,
`Processing, Vol. 83, pp.2641-2673.
`[3] L.C. Parra, Steerable Frequency-invariant Beamforming for
`Aribitrary arrays, J. Acoust. Soc. Am, 199(6), pp.3839-3847,
`June 2006.
`[4] M. Brandstein and D. Ward, Microphone Arrays. Springer,
`2001.
`[5] D.B. Ward, R.A. Kennedy and R.C. Williamson, Constant
`Directivity Beamforming, Microphone Arrays, pp3-17.
`[6] S. Yan and Y. Ma, Design of FIR Beamformer with Frequency
`Invariant Pattern via Jointly Optimizing Spatial and Frequency
`Response, ICASSP 2005 pp.789-792.
`[7] I. Tashev and H.S. Malvar, A new beamformer design
`algorithm for microphone arrays. Proceedings of International
`Conference of Acoustic, Speech and Signal Processing ICASSP
`2005, Philadelphia, PA, USA, March 2005.
`[8] M. Omologo, M. Matassoni and P. Svaizer, Speech
`Recognition with Microphone Array, Microphone Arrays,
`pp.331-34.
`
`Amazon Ex. 1008, Page 9 of 9
`
`