Abstract: A sound source model storage section stores a sound source model that represents an audio signal emitted from a sound source in the form of a probability density function. An observation signal, which is obtained by collecting the audio signal, is converted into a plurality of frequency-specific observation signals each corresponding to one of a plurality of frequency bands. Then, a dereverberation filter corresponding to each frequency band is estimated by using the frequency-specific observation signal for the frequency band on the basis of the sound source model and a reverberation model that represents a relationship for each frequency band among the audio signal, the observation signal and the dereverberation filter. A frequency-specific target signal corresponding to each frequency band is determined by applying the dereverberation filter for the frequency band to the frequency-specific observation signal for the frequency band, and the resulting frequency-specific target signals are integrated.

Abstract: Communication systems are described, including both portable handset and headset devices, which use a number of microphone configurations to receive acoustic signals of an environment. The microphone configurations include, for example, a two-microphone array including two unidirectional microphones, and a two-microphone array including one unidirectional microphone and one omnidirectional microphone. The communication systems also include Voice Activity Detection (VAD) devices to provide information of human voicing activity. Components of the communications systems receive the acoustic signals and voice activity signals and, in response, automatically generate control signals from data of the voice activity signals. Components of the communication systems use the control signals to automatically select a denoising method appropriate to data of frequency subbands of the acoustic signals.

Abstract: Various embodiments reduce noise within a particular environment, while isolating and capturing speech in a manner that allows operation within an otherwise noisy environment. In one embodiment, an array of one or more microphones is used to selectively eliminate noise emanating from known, generally fixed locations, and pass signals from a pre-specified region or regions with reduced distortion.

Abstract: A sound processor includes a conversion unit converts a reference sound signal corresponding to a base of sound to be output and an observation sound signal based on each of sound signals output by a plurality of sound receiving units into frequency components, an echo suppression unit estimates echo derived from sound based on a converted reference sound signal and suppressing the estimated echo in a converted observation sound signal, a noise suppression unit estimates noise based on an arrival direction of sound and suppressing the estimated noise in the converted observation sound signal and an integrating process unit suppresses, with respect to each frequency component, echo and noise in the converted sound signal based on a observation sound signal obtained after echo suppression and a observation sound signal obtained after noise suppression.

Abstract: A method and apparatus are provided for generating a noise reduced output signal from sound received by a first microphone. The method includes transforming the sound received by the first microphone into a first input signal and transforming sound received by a second microphone into a second input signal. The method includes calculating, for each of a plurality of frequency components, an energy transfer function value as a real-valued quotient by dividing a temporally averaged product of an amplitude of the first input signal and the second input signal by a temporally averaged absolute square of the second input signal, calculating a gain value as a function of the calculated energy transfer function value, and generating the noise reduced output signal based on the product of the first input signal and the calculated gain value at each of the plurality of frequency components.

Abstract: A system improves speech detection or processing by identifying registration signals. The system encodes a limited frequency band by varying the amplitude of a pulse width modulated signal between predefined values. The signal is separated into frequency bins that identify amplitude and phase. The registration signal is measured by comparing a difference in average acoustic power in a plurality of adjacent bins over time.

Abstract: In a system for estimating the power spectral density of acoustical background noise when the level of a smoothed power spectral density signal increases, an increment value is increased, starting from a minimum increment value, by a predetermined amount until a maximum increment value is reached if at the same time the value of the power spectral density currently determined in a new calculation cycle is larger than the estimate value of the power spectral density of the background noise determined in the previous calculation cycle. For cases in which the level of the smoothed power spectral density decreases, the amplitude of the decrement value is increased, starting from a minimum decrement value, by a predetermined amount until a maximum decrement value is reached if at the same time the value of the power spectral density currently determined in a new calculation cycle is smaller than the estimate value of the power spectral density of the background noise determined in the previous calculation cycle.

Abstract: A signal processing apparatus is configured to change volume level or frequency characteristics of an input signal with a limited bandwidth in a first frequency range. The apparatus includes: an information extracting unit configured to extract second frequency characteristic information from a collection signal with a limited bandwidth in a second frequency range different from the first frequency range; a frequency characteristic information extending unit configured to estimate first frequency characteristic information from the second frequency characteristic information extracted by the information extracting unit, the first frequency characteristic information including the first frequency range; and a signal correcting unit configured to change volume level or frequency characteristics of the input signal according to the first frequency characteristic information obtained by the frequency characteristic information extending unit.

Abstract: A method for processing signals for reducing noise components in a vehicular microphone system (500A) compares (517) a plurality of noise component values (511, 513, 515) for at least one frequency band of plurality of frequency bands. At least one frequency band is then downwardly expanded (521) by a predetermined expansion ratio (523) for providing a noise reduced signal (535) when determining a value is below a predetermined threshold (519).

Abstract: A speech enhancement device and a method for the same are included. The device includes a down-converter, a speech enhancement processor, and an up-converter. The method includes steps of down-converting audio signals to generate down-converted audio signals; performing speech enhancement on the down-converted audio signals to generate speech-enhanced audio signals; and up-converting the speech enhancement audio signals to generate up-converted audio signals.

Abstract: A method for effecting a machine-based determination of speech intelligibility in an aircraft during flight operations includes: (a) in no particular order: (1) providing a representation of a machine-based speech evaluating signal; and (2) providing a representation of in-flight noise; (b) combining the representation of a machine-based speech evaluation signal and the representation of in-flight noise to obtain a combined noise signal; and (c) employing the combined noise signal to present the machine-based determination of speech intelligibility in an aircraft during flight operations.

Abstract: An active noise reduction system is provided for receiving an audio input signal and a noise interference signal and calculating an audio broadcasting signal according to a Feedback Filtered-X Least-Mean-Square (FFXLMS) algorithm, wherein the FFXLMS algorithm optimizes a (convergence factor) ? so as to decrease the numbers of divisions operated by the active noise reduction system and increase the operation speed of the active noise reduction system.

Abstract: An acoustic shock protection method and device are provided. A pattern analysis-based approach is taken to an input signal to perform feature extraction. A parameter space is identified, which is corresponding to, the signal space of the input signal. A rule-based decision approach is taken to the parameter space to detect an acoustic shock event. The device may be advantageously implemented using a weighted overlap-add approach to provide low group delay, high-fidelity and a high degree of protection from acoustic shock events.

Abstract: A method and device for transforming ambient audio are provided. Example embodiments may include monitoring ambient audio proximate to a sound processing device located in an environment. The device may receive a selection from a user interface. The selection may comprise one of a number of available first selections, each available first selection identifying one of multiple transformation modes. The device may access memory to obtain transformation audio and process the transformation audio, based on the ambient audio and the selection. The device may also use the transformation audio to provide modified output audio for propagation into the environment.

Abstract: This invention proposed an Echo Suppressor which can efficiently suppress both echoes and background noise without introducing “choppiness”. The Echo Suppressor System includes said two adaptive gains Gr(RSR) and Gn(NSR), said one adaptive zeros-filter A1(z) and said one adaptive poles-filter A2(z); wherein, thr gain Gr(RSR) is controlled by RSR (Residual echo level to Signal level Ratio); the gain Gn(NSR) is controlled by NSR (Noise signal level to current Signal (Tx) level Ratio); the filter A1(z) is converted from LSF1 obtained from the first modification of LSFTx (Line Spectral Frequencies of Tx signal); the filter A2(z) is converted from LSF2 obtained from the second modification of LSFTx.

Abstract: A method for separating a sound source from a mixed signal, includes Transforming a mixed signal to channel signals in frequency domain; and grouping several frequency bands for each channel signal to form frequency clusters. Further, the method for separating the sound source from the mixed signal includes separating the frequency clusters by applying a blind source separation to signals in frequency domain for each frequency cluster; and integrating the spectrums of the separated signal to restore the sound source in a time domain wherein each of the separated signals expresses one sound source.

Abstract: In one embodiment, an automated interferometric noise measurement system includes: a signal source adapted to provide a carrier signal; a delay line adapted to delay a first version of the carrier signal to provide a delayed signal to a device-under-test (DUT); a variable attenuator adapted to attenuate a second version of the carrier signal to provide an attenuated signal; a first variable phase-shifter adapted to phase-shift the attenuated signal to provide a first phase-shifted signal; a hybrid coupler adapted to receive an output signal from the DUT and the first phase-shifted signal to provide a carrier-suppressed signal and a carrier-enhanced signal; a low-noise amplifier adapted to amplify the carrier-suppressed signal to provide an amplified signal; a second variable phase-shifter adapted to phase-shift a version of the carrier-enhanced signal to provide a second phase-shifted signal; a first mixer adapted to mix a first version of the amplified signal and the second phase-shifted signal to provide a

Abstract: An audio signal processing method for processing input audio signals of plural channels includes calculating at least one feature quantity representing a difference between channels of input audio signals, selecting at least one weighting factor according to the feature quantity from at least one weighting factor dictionary prepared by learning beforehand, and subjecting the input audio signals of plural channels to signal processing including noise suppression and weighting addition using the selected weighting factor to generate output an output audio signal.

Abstract: A system estimates the spectral noise power density of an audio signal includes a spectral noise power density estimation unit, a correction term processor, and a combination processor. The spectral noise power density estimation unit may provide a first estimate of the spectral noise power density of the audio signal. The correction term processor may provide a time dependent correction term based, at least in part, on a spectral noise power density estimation error of the actual spectral noise power density. The correction term may be determined so that the spectral noise power density estimation error is reduced. The combination processor may combine the first estimate with the correction term to obtain a second estimate of the spectral noise power density that may be used for subsequent signal processing to enhance a desired signal component of the audio signal.

Abstract: A method for determining coefficients of a family of cascaded second order Infinite Impulse Response (IIR) parametric filters used for equalizing a room response. The method includes determining parameters of each IIR parametric filter from poles or roots of a reasonably high-order Linear Predictive Coding (LPC) model. The LPC model is able to accurately model the low-frequency room response modes providing better equalization of loudspeaker and room acoustics, particularly at the low frequencies. Advantages of the method include fast and efficient computation of the LPC model using a Levinson-Durbin recursion to solve the normal equations that arise from the least squares formulation. Due to possible band interactions between the cascaded IIR parametric filters, the method further includes optimizing the Q value of each filter to better equalize the room response.

Abstract: Systems and methods for envelope-based acoustic echo cancellation in a communication device are provided. In exemplary embodiments, a primary acoustic signal is received via a microphone of the communication device, and a far-end signal is received via a receiver. Frequency analysis is performed on the primary acoustic signal and the far-end acoustic signal to obtain frequency sub-bands. An echo gain mask based on magnitude envelopes of the primary and far-end acoustic signals for each frequency sub-band is generated. A noise gain mask based on at least the primary acoustic signal for each frequency sub-band may also be generated. A combination of the echo gain mask and noise gain mask may then be applied to the primary acoustic signal to generate a masked signal. The masked signal is then output.

Abstract: The present system proposes a technique called the spectro-temporal varying technique, to compute the suppression gain. This method is motivated by the perceptual properties of human auditory system; specifically, that the human ear has higher frequency resolution in the lower frequencies band and less frequency resolution in the higher frequencies, and also that the important speech information in the high frequencies are consonants which usually have random noise spectral shape. A second property of the human auditory system is that the human ear has lower temporal resolution in the lower frequencies and higher temporal resolution in the higher frequencies. Based on that, the system uses a spectro-temporal varying method which introduces the concept of frequency-smoothing by modifying the estimation of the a posteriori SNR. In addition, the system also makes the a priori SNR time-smoothing factor depend on frequency.

Abstract: A band separating unit 5 carries out a band division of a plurality of power spectra into which an input signal is converted by a time-to-frequency converting unit 2 to combine power spectra into each subband, and a band representative component generating unit 6 defines a power spectrum having a maximum among the plurality of power spectra within each subband as a representative power spectrum. A noise suppression amount generating unit 7 calculates an amount of noise suppression for each subband by using the representative power spectrum and a noise spectrum, and a noise suppressing unit 9 suppresses the amplitudes of the power spectra according to the amount of noise suppression.

Abstract: An exemplary method of dynamically adjusting an amount of noise reduction applied in an auditory prosthesis system includes dividing an audio signal presented to a patient into a plurality of analysis channels each containing a signal representative of a distinct frequency portion of the audio signal, determining an overall noise level of the signals within the analysis channels, and dynamically adjusting an amount of noise reduction applied to the signals within the analysis channels in accordance with the determined overall noise level. The dynamic adjustment of noise reduction is configured to minimize the amount of noise reduction applied to the signals within the analysis channels if the overall noise level is less than a predetermined minimum threshold. Corresponding methods and systems are also disclosed.

Abstract: Systems and methods for utilizing inter-microphone level differences to attenuate noise and enhance speech are provided. In exemplary embodiments, energy estimates of acoustic signals received by a primary microphone and a secondary microphone are determined in order to determine an inter-microphone level difference (ILD). This ILD in combination with a noise estimate based only on a primary microphone acoustic signal allow a filter estimate to be derived. In some embodiments, the derived filter estimate may be smoothed. The filter estimate is then applied to the acoustic signal from the primary microphone to generate a speech estimate.

Abstract: A first matrix (W(k)) indicating frequency characteristics of a separation filter is calculated from input signals of a plurality of channels. A second matrix (Ws(k)) is calculated by using the restriction coefficients (Ci(k)) for restricting the separation filter and the first matrix, and separation filter coefficients (wsij(s)) are calculated by using the second matrix. With use of the separation filter coefficients, separation signals (ysi(t)) are then calculated from the input signals. A third matrix (Ws?1(k)) is then calculated by transforming the second matrix into an inverse matrix at each frequency, and reproduction filter coefficients (a?I1(s), a?I2(s)) are calculated by using the third matrix. With use of the reproduction filter coefficients, the synthesized signal of each channel is calculated by using the separation signals.

Abstract: An audio signal is conveyed more efficiently by transmitting or recording a baseband of the signal with an estimated spectral envelope and a noise-blending parameter derived from a measure of the signal's noise-like quality. The signal is reconstructed by translating spectral components of the baseband signal to frequencies outside the baseband, adjusting phase of the regenerated components to maintain phase coherency, adjusting spectral shape according to the estimated spectral envelope, and adding noise according to the noise-blending parameter. Preferably, the transmitted or recorded signal also includes an estimated temporal envelope that is used to adjust the temporal shape of the reconstructed signal.

Abstract: Provided is a method and apparatus for encoding/decoding an audio signal. Sections which are not used to output noise components near important spectral components and sub-bands which are not used to output noise components, are determined to be encoded or decoded, so that the efficiency of encoding and decoding an audio signal increases, and sound quality can be improved using less bits.

Abstract: A noise reducing apparatus includes: a voice signal inputting unit inputting an input voice signal; a noise occurrence period detecting unit detecting a noise occurrence period; a noise removing unit removing a noise for the noise occurrence period; a generation source signal acquiring unit acquiring a generation source signal with a time duration corresponding to a time duration corresponding to the noise occurrence period; a pitch calculating unit calculating a pitch of an input voice signal interval; an interval signal setting unit setting interval signals divided in each unit period interval; an interpolation signal generating unit generating an interpolation signal with the time duration corresponding to the noise occurrence period and alternately arranging the interval signal in a forward time direction and the interval signal in a backward time direction; and a combining unit combining the interpolation signal and the input voice signal, from which the noise is removed.

Abstract: A noise reduction processing apparatus includes a timing signal detection unit that detects an operation timing signal indicating a timing when an operation unit is operated, an audio signal acquisition unit that acquires an audio signal, and on the basis of the operation timing signal, a noise reduction processing unit that calculates first frequency spectra of an audio signal acquired during a time period which has high possibility that noise caused by an operation of the operation unit is generated and second frequency spectra of an audio signal acquired during a time period which has high possibility that the noise is not generated, and calculates a noise reduction audio signal obtained by performing noise reduction to the audio signal on the basis of frequency spectra in which at least a part of the calculated first frequency spectra are replaced with corresponding portions of the calculated second frequency spectra.

Abstract: A noise reducing device includes: a sound-signal input unit that inputs a sound signal; a time-to-frequency converting unit that converts the input sound signal obtained by being input by the sound-signal input unit into a frequency signal; a patterning unit that calculates, for each of divided frequencies of the frequency signal, an nth order polynomial (n is a natural number) as a polynomial for interpolating sampling points and acquires a coefficient pattern including a set of values of coefficients of respective orders of the polynomial; a matching-data storing unit that stores, in association with the divided frequencies, matching data indicating a matching range as a range of a coefficient pattern regarded as noise; and a noise determining unit that determines, on the basis of a result obtained by comparing the coefficient pattern acquired by the patterning unit and the matching range indicated by the matching data, at least presence or absence of noise occurrence at a divided frequency corresponding to

Abstract: An interference signal removing apparatus of a radio frequency (RF) receiver includes a low noise amplification unit which performs low noise amplification, a feedback processing unit which removes a necessary signal in a desired band from a signal output from the low noise amplification unit, and performs feedback of the signal from which the necessary signal is removed, and a signal processing unit which transmits a processed RF signal by synthesizing an input RF signal and the feedback signal to the noise amplification unit.

Abstract: A howling suppression apparatus suppresses a howling caused in an acoustic system including a sound collection device and a sound emission device. An estimation part generates an estimated signal by estimating a feedback sound reaching the sound collection device from the sound emission device. An adjustment part generates an estimated signal by adjusting the estimated signal. A spectrum subtraction part generates an acoustic signal using a result of subtracting a frequency spectrum of the estimated signal from a frequency spectrum of an acoustic signal. A filter part generates an acoustic signal by suppressing a component of a frequency band including a howling frequency F among the acoustic signal. An acoustic signal in which the acoustic signal is amplified by an amplifier is supplied to the sound emission device.

Abstract: An audio signal processing apparatus includes: a dividing section dividing each of audio signals of a plurality of channels into a plurality of frequency bands; a phase difference calculating section calculating a phase difference between the audio signals of the plurality of channels, for each of the plurality of frequency bands divided by the dividing section; a level ratio calculating section calculating a level ratio between the audio signals of the plurality of channels, for each of the plurality of frequency bands divided by the dividing section; and an audio signal processing section performing output gain setting with respect to divided signals obtained by the dividing section, on the basis of the phase difference and the level ratio for each of the plurality of frequency bands calculated by the phase difference calculating section and the level ratio calculating section.

Abstract: An automatic gain controller and a method using the same are provided. The automatic gain controller and method analyze background noise by operating a microphone mounted in a mobile communication terminal and automatically control the gain of the signal part which is non-audible due to the background noise. Thus, a user may listen to music by using an earphone or a headphone connected to the mobile communication terminal in an environment with background noise. The method includes receiving an audio signal to be reproduced, receiving a background noise signal introduced through a microphone, controlling a gain of the audio signal by comparing the background noise signal and the audio signal and outputting the gain-controlled audio signal so that the user can listen to the gain-controlled audio signal.

Abstract: Provided are an apparatus and method for preventing noise. The apparatus estimates a noise signal from a signal transformed into a frequency domain, uses the estimated noise signal to estimate the amplitude of the frequency-domain signal according to a frequency band, and then calculates a phase difference according to a frequency band and eliminates or prevents noise from the amplitude-estimated frequency-domain signal based on the calculated phase difference according frequency band.

Abstract: A noise cancellation apparatus includes a noise estimation module for receiving a noise-containing input speech, and estimating a noise therefrom to output the estimated noise; a first Wiener filter module for receiving the input speech, and applying a first Wiener filter thereto to output a first estimation of clean speech; a database for storing data of a Gaussian mixture model for modeling clean speech; and an MMSE estimation module for receiving the first estimation of clean speech and the data of the Gaussian mixture model to output a second estimation of clean speech. The apparatus further includes a final clean speech estimation module for receiving the second estimation of clean speech from the MMSE estimation module and the estimated noise from the noise estimation module, and obtaining a final Wiener filter gain therefrom to output a final estimation of clean speech by applying the final Wiener filter gain.

Abstract: A signal processing system detects reverberation. The system may suppress the reverberation and improve signal quality. The system analyzes frequency bands of an input signal to determine whether reverberation characteristics are present. When reverberation is detected, the system may attenuate the reverberant frequency band to reduce or eliminate the reverberation.

Abstract: An apparatus for processing an audio signal and method thereof are disclosed, by which a local dynamic range of an audio signal can be adaptively normalized as well as a maximum dynamic range of the audio signal. The present invention includes receiving, by an audio processing apparatus, a signal, and feedback information estimated based on a normalizing gain; generating a noise estimation based on the signal; computing a gain filter for noise canceling, based on the noise estimation and the signal; and, obtaining a restricted gain filter by applying the feedback information to the gain filter.

Abstract: An apparatus for processing an audio signal and method thereof are disclosed, by which a local dynamic range of an audio signal can be adaptively normalized as well as a maximum dynamic range of the audio signal. The present invention includes receiving a signal, by an audio processing apparatus; computing a long-term power and a short-term power by estimating power of the signal; generating a slow gain based on the long-term power; generating a fast gain based on the short-term power; obtaining a final gain by combining the slow gain and the fast gain; and, modifying the signal using the final gain.

Abstract: According to an aspect of the invention, there is provided a noise suppressing apparatus comprising: a fifth unit configured to calculate a gain for noise suppression, based on the first signal-to-noise ratio for each frequency band and the second signal-to-noise ratio for an entire frequency band; an eighth unit configured to calculate an upper limit value of a noise suppression amount for each frequency band, based on the second signal-to-noise ratio; a ninth unit configured to calculate the noise suppression amount for each frequency band, based on the first signal-to-noise ratio; and a tenth unit configured to limit, based on the upper limit value, the noise suppression amount so as to calculate the gain.

Abstract: A system and method for analyzing a signal to monitor the dynamics of its magnitude and frequency characteristics over time. An electronic circuit for identifying feedback in an audio signal, formed in accordance with embodiments of the invention may comprise a feedback control block operable to determine a candidate frequency having potential feedback such that the feedback control block is further operable to perform an iterative analysis of the magnitude of the audio signal at the candidate frequency to determine the growth characteristics of the signal. The electronic circuit may further include a test filter block operable to deploy a test filter at a candidate frequency and a permanent filter block operable to deploy a permanent filter at the candidate frequency if the feedback control block determines that the growth characteristics of the signal at the candidate frequency comprises feedback characteristics after the test filter has been deployed.

Abstract: Provided is a method and apparatus for encoding/decoding an audio signal. Sections which are not used to output noise components near important spectral components and sub-bands which are not used to output noise components, are determined to be encoded or decoded, so that the efficiency of encoding and decoding an audio signal increases, and sound quality can be improved using less bits.

Abstract: In one embodiment, one or more users may be participating in a conversation. In one example, a first user may be speaking into a speaker end device and a second user may be listening at a listener end device. The second user may be in an environment where noise may be present. Particular embodiments determine characteristics of the noise at the listener end device. Characteristics of a voice signature for a user speaking with the speaker end device are also determined. Comprehension enhancement of voice signals received from speaker end device is then performed based on characteristics of the noise at the listener end device and characteristics of the voice signature. For example, the signature of the voice signals may be altered to lessen the overlap with the noise.

Abstract: A method for reducing noise in a vehicle by adjusting the rotational speed of various rotational components.

Abstract: An oscillation-echo preventing circuit has a microphone/speaker unit and a voltage-canceling circuit for canceling voltages of audio receive signals. The microphone/speaker unit has a main body, at least two microphones, and a speaker or an earphone. The microphone seals a first inside space from an outside space. The microphone seals the first inside space from a second inside space. The speaker or the earphone seals the first inside space from the outside space. The voltage-canceling circuit cancels out the voltages of audio receive signals coming from the microphones, respectively, generating an output of minimum magnitude. Thus, the circuit can sufficiently suppress oscillation and echoing.

Abstract: The present proposes new methods and an apparatus for enhancement of source coding systems utilising high frequency reconstruction (HFR). It addresses the problem of insufficient noise contents in a reconstructed highband, by Adaptive Noise-floor Addition. It also introduces new methods for enhanced performance by means of limiting unwanted noise, interpolation and smoothing of envelope adjustment amplification factors. The present invention is applicable to both speech coding and natural audio coding systems.

Abstract: A sound signal correcting apparatus converts an acquired sound signal into a phase spectrum and an amplitude spectrum by an FFT process, compares the amplitude spectrum of the obtained sound signal with a noise model so that a correction coefficient used for correcting the amplitude spectrum of the sound signal is derived, smoothes waveform of the amplitude spectrum of the sound signal using the derived correction coefficient, and converts the sound signal into a sound signal where the amplitude spectrum is corrected by performing an inverse FFT process on the phase spectrum and the smoothed amplitude spectrum.

Abstract: A sound input device includes a differential microphone, configured to receive sound including noise, and generate a first signal in accordance with the sound; a detector, configured to detect the noise, and generate a second signal in accordance with the detected noise; and a controller, configured to control at least one of suppression of high-frequency components of the first signal and changing of a frequency band to be suppressed of the first signal based on the second signal.

Abstract: A microphone-array-based speech recognition system using a blind source separation (BBS) and a target speech extraction method in the system are provided. The speech recognition system performs an independent component analysis (ICA) to separate mixed signals input through a plurality of microphone into sound-source signals, extracts one target speech spoken for speech recognition from the separated sound-source signals by using a Gaussian mixture model (GMM) or a hidden Markov Model (HMM), and automatically recognizes a desired speech from the extracted target speech. Accordingly, it is possible to obtain a high speech recognition rate even in a noise environment.