Abstract: A noise reduction apparatus includes an analysis unit for converting input into a signal of a frequency area, a suppression unit for suppressing the signal, and a synthesis unit for synthesizing a signal of a time area. The apparatus further includes an estimation unit for estimating, using the output of the analysis unit, information corresponding to at least pure voice element excluding noise element in an input voice signal as voice information which is the basic voice information for calculation of a suppression gain of a signal, and a unit for calculating a suppression gain corresponding to the output of the estimation unit and the analysis unit and providing it for the suppression unit.

Abstract: Computer implemented speech processing is disclosed. First and second voice segments are extracted from first and second microphone signals originating from first and second microphones. The first and second voice segments correspond to a voice sound originating from a common source. An estimated source location is generated based on a relative energy of the first and second voice segments and/or a correlation of the first and second voice segments. A determination whether the voice segment is desired or undesired may be made based on the estimated source location.

Abstract: A method and system for automatic gain control of a speech signal in a communication system are disclosed. The gain of the speech signal can be controlled, based on a calculated gain value. This gain value is calculated on the basis of energy calculation and speech activity identification in the speech signal which is done by means of the encoder. Encoding the gain controlled speech signal for transmission follows the step of gain control.

Abstract: Various embodiments are disclosed relating to the real-time monitoring and control for audio devices. An apparatus may include a peripheral audio device configured to operate in an operational mode or a debug mode, the peripheral audio device including an audio enhancement logic configured to include at least one tunable parameter. The apparatus may also include the peripheral audio device being further configured to transmit and receive data via a data channel to allow a debug or test to be performed on the peripheral audio device, while operating in the debug mode, and the at least one tunable parameter to be adjusted.

Abstract: Systems and methods for receiving natural language queries and/or commands and execute the queries and/or commands. The systems and methods overcome the deficiencies of prior art speech query and response systems through the application of a complete speech-based information query, retrieval, presentation and command environment. This environment makes significant use of context, prior information, domain knowledge, and user specific profile data to achieve a natural environment for one or more users making queries or commands in multiple domains. Through this integrated approach, a complete speech-based natural language query and response environment can be created. The systems and methods creates, stores and uses extensive personal profile information for each user, thereby improving the reliability of determining the context and presenting the expected results for a particular question or command.

Abstract: The present invention discloses an audio signal segmentation algorithm comprising the following steps. First, an audio signal is provided. Then, an audio activity detection (AAD) step is applied to divide the audio signal into at least one noise segment and at least one noisy audio segment. Then, an audio feature extraction step is used on the noisy audio segment to obtain multiple audio features. Then, a smoothing step is applied. Then, multiple speech frames and multiple music frames are discriminated. The speech frames and the music frames compose at least one speech segment and at least one music segment. Finally, the speech segment and the music segment are segmented from the noisy audio segment.

Abstract: A signal processing system which discriminates between voice signals and data signals modulated by a voiceband carrier. The signal processing system includes a voice exchange, a data exchange and a call discriminator. The voice exchange is capable of exchanging voice signals between a switched circuit network and a packet based network. The signal processing system also includes a data exchange capable of exchanging data signals modulated by a voiceband carrier on the switched circuit network with unmodulated data signal packets on the packet based network. The data exchange is performed by demodulating data signals from the switched circuit network for transmission on the packet based network, and modulating data signal packets from the packet based network for transmission on the switched circuit network. The call discriminator is used to selectively enable the voice exchange and data exchange.

Abstract: A method and apparatus are provided for using the uncertainty of a noise-removal process during pattern recognition. In particular, noise is removed from a representation of a portion of a noisy signal to produce a representation of a cleaned signal. In the meantime, an uncertainty associated with the noise removal is computed and is used with the representation of the cleaned signal to modify a probability for a phonetic state in the recognition system. In particular embodiments, the uncertainty is used to modify a probability distribution, by increasing the variance in each Gaussian distribution by the amount equal to the estimated variance of the cleaned signal, which is used in decoding the phonetic state sequence in a pattern recognition task.

Abstract: An echo suppressing system includes: a sound output device for outputting sound based on a sound signal, including a passing section for allowing passage of a component of a different frequency band, and a plurality of sound output sections, each of which outputs sound based on each of the plurality of sound signals passed through the passing section; a summer for summing the plurality of sound signals to generate a reference sound signal; a sound input device for converting input sound into a sound signal; and an echo suppressor for suppressing echo based on the sound output by the sound output device, including an input section to which a sound signal is input from the sound input device as an observation sound signal, and a correction section for correcting the observation sound signal so as to suppress echo included in the observation sound signal.

Abstract: An audio encoding device which can improve encoding performance while performing division search on an algebraic codebook in an audio encoding. In a distortion minimizing unit (112) of a CELP encoding device: a maximum correlation value calculation unit (221) calculates a correlation value by using each pulse and a target signal in each candidate position for four pulses constituting the fixed codebook so as to acquire a maximum value of the correlation value for each pulse and calculates a maximum correlation value by using the maximum value of the correlation value; a sorting unit (222) divides the four pulses into two subsets each having two pulses; and a search unit (224) performs a division search on the fixed codebook and acquires a code indicating the positions and polarities of the four pulses where the encoding distortion is minimum.

Abstract: Architecture that employs a combination of in-band signaling (e.g., DTMF) with speech recognition to deliver usability improvements. The in-band signaling allows the user to indicate to the system when a barge-in operation is occurring and/or when to start listening to subsequent speech input and optionally, when to stop listening for further speech input. The in-band signaling can be utilized during a telephone call and using wireline and wireless telephones. Moreover, the architecture can be incorporated at the platform level requiring little, if any, application changes to support the new mode of operation.

Abstract: A method and apparatus to disturb a voice signal by attenuating and masking the voice signal are provided. The method includes; receiving a voice signal from a wired or wireless network; obtaining a masked voice signal by dividing the received voice signal into a plurality of segments of the same size; outputting the received voice signal and receiving a feedback signal of the output voice signal; obtaining an attenuated voice signal by performing a first sound attenuation operation on the feedback signal; and combining the attenuated voice signal and the masked voice signal and outputting the result of the combination as disturbing sound.

Abstract: A speech distinction method, which includes dividing an input voice signal into a plurality of frames, obtaining parameters from the divided frames, modeling a probability density function of a feature vector in state j for each frame using the obtained parameters, and obtaining a probability P0 that a corresponding frame will be a noise frame and a probability P1 that the corresponding frame will be a speech frame from the modeled PDF and obtained parameters. Further, a hypothesis test is performed to determine whether the corresponding frame is a noise frame or speech frame using the obtained probabilities P0 and P1.

Abstract: The invention regards a method for processing audio-signals whereby audio signals are captured at two spaced apart locations and subject to a transformation in the perceptual domain (Bar or Mel), whereupon: a) a (blind or supervised) source separation process is performed to give a first estimate of the wanted signal parts and the noise parts of the microphone signals and b) a coherence based separation process is performed to give a second estimate of the wanted signal parts and the noise parts of the microphone signals, and where further a sound field diffuseness detection is performed on the at least two signals, whereby further the sound field diffuseness detections is used to mix the output from the blind source separation and the coherence based separation process in order to achieve the best possible signal.

Abstract: A method for enhancing speech includes extracting a center channel of an audio signal, flattening the spectrum of the center channel, and mixing the flattened speech channel with the audio signal, thereby enhancing any speech in the audio signal. Also disclosed are a method for extracting a center channel of sound from an audio signal with multiple channels, a method for flattening the spectrum of an audio signal, and a method for detecting speech in an audio signal. Also disclosed is a speech enhancer that includes a center-channel extract, a spectral flattener, a speech-confidence generator, and a mixer for mixing the flattened speech channel with original audio signal proportionate to the confidence of having detected speech, thereby enhancing any speech in the audio signal.

Abstract: An apparatus and a method thereof, processes a voice signal of a mobile terminal in a mobile communication system. The apparatus to process a received-voice signal received through a wireless channel in a mobile terminal includes a digital signal processing unit to generate an encoded packet and frame type information defining a characteristic of the encoded packet by performing voice encoding on an audible signal input from a microphone. The apparatus also includes a received-voice controlling unit to determine a noise level in consideration of the frame type information and a level of the audible signal, and to control at least one of a tone and a volume of received voice by the determined noise level.

Abstract: A method for determining unbiased signal amplitude estimates () after cepstral variance modification of a discrete time domain signal (s(t)), wherein the cepstrally-modified spectral amplitudes () of the discrete time domain signal (s(t)) are X-distributed with 2{tilde over (?)} degrees of freedom. A bias reduction factor (r) is determined using the equation r 2 = ? ? ~ ? ? ? ? ( ? ~ ) - ? ? ( ? ) , where 2? are the degrees of freedom of the X-distributed spectral amplitudes of the discrete time domain signal (s(t)) and ? ? ( x ) = - 0.5772 - ? n = 0 ? ? ( 1 x + n - 1 1 + n ) ; then the unbiased signal amplitude estimates () are determined by multiplying the cepstrally-modified spectral amplitudes () with the bias reduction factor (r) according to the equation =.

Abstract: A voice/music determining apparatus is configured to calculate first feature parameters for discriminating between a voice signal and a musical signal; and calculate second feature parameters for discriminating between a musical signal and a background-sound-superimposed voice signal. A first score is calculated to indicate likelihood that the input audio signal is a voice signal or a musical signal as a sum of weight-multiplied first feature parameters. A second score is calculated to indicate likelihood that the input audio signal is a musical signal or a background-sound-superimposed voice signal as a sum of weight-multiplied second feature parameters. It is determined whether the input audio signal is a voice signal or a musical signal on the basis of the first score. Further, it is determined whether the musical signal is the input audio signal is a background-sound-superimposed voice signal on the basis of the second score.

Abstract: Apparatus, method, and medium for processing an audio signal using a correlation between bands are provided. The apparatus includes an encoding unit encoding an input audio signal and a decoding unit decoding the encoded input audio signal.

Abstract: A method, system and computer program for encoding speech according to a source-filter model. The method comprises deriving a spectral envelope signal representative of a modelled filter and a first remaining signal representative of a modelled source signal, and deriving a second remaining signal from the first remaining signal by, at intervals during the encoding: exploiting a correlation between approximately periodic portions in the first remaining signal to generate a predicted version of a later portion from a stored version of an earlier portion, and using the predicted-version of the later portion to remove an effect of said periodicity from the first remaining signal. The method further comprises, once every number of intervals, transforming the stored version of the earlier portion of the first remaining signal prior to generating the predicted version of the respective later portion.

Abstract: A method of filtering a speech signal for speech encoding in a communications network, includes determining a cut off frequency for a filter, wherein a component of the speech signal in a frequency range less than the cut off frequency is to be attenuated by the filter; receiving the speech signal at the filter; determining at least one parameter of the received speech signal, the at least one parameter providing an indication of the energy of the component of the received speech signal that is to be attenuated; and adjusting the cut off frequency in dependence on the at least one parameter, thereby adjusting the frequency range to be attenuated.

Abstract: A signal processing system according to various aspects of the present invention includes an excursion signal generator, a scaling system and a filter system. The excursion signal generator identifies a peak portion of a signal that exceeds a threshold and generates a corresponding excursion signal. The scaling system applies a real scale factor to contiguous sets of excursion samples in order to optimize peak-reduction performance. The filter system filters the excursion signal to remove unwanted frequency components from the excursion signal. The filtered excursion signal may then be subtracted from a delayed version of the original signal to reduce the peak. The signal processing system may also control power consumption by adjusting the threshold. The signal processing system may additionally adjust the scale of the excursion signal and/or individual channel signals, such as to meet constraints on channel noise and output spectrum, or to optimize peak reduction.

Abstract: Methods and apparatus to operate an audience metering device with voice commands are described herein. An example method to identify audience members based on voice, includes: obtaining an audio input signal including a program audio signal and a human voice signal; receiving an audio line signal from an audio output line of a monitored media device; processing the audio line signal with a filter having adaptive weights to generate a delayed and attenuated line signal; subtracting the delayed and attenuated line signal from the audio input signal to develop a residual audio signal; identifying a person that spoke to create the human voice signal based on the residual audio signal; and logging an identity of the person as an audience member.

Abstract: A high quality speech is reproduced with a small data amount in speech coding and decoding for performing compression coding and decoding of a speech signal to a digital signal. In speech coding method according to a code-excited linear prediction (CELP) speech coding, a noise level of a speech in a concerning coding period is evaluated by using a code or coding result of at least one of spectrum information, power information, and pitch information, and various excitation codebooks are used based on an evaluation result.

Abstract: A high quality speech is reproduced with a small data amount in speech coding and decoding for performing compression coding and decoding of a speech signal to a digital signal. In speech coding method according to a code-excited linear prediction (CELP) speech coding, a noise level of a speech in a concerning coding period is evaluated by using a code or coding result of at least one of spectrum information, power information, and pitch information, and various excitation codebooks are used based on an evaluation result.

Abstract: A high quality speech is reproduced with a small data amount in speech coding and decoding for performing compression coding and decoding of a speech signal to a digital signal. In speech coding method according to a code-excited linear prediction (CELP) speech coding, a noise level of a speech in a concerning coding period is evaluated by using a code or coding result of at least one of spectrum information, power information, and pitch information, and various excitation codebooks are used based on an evaluation result.

Abstract: A noise detection apparatus includes a time-frequency transform unit configured to transform an input signal from a time domain to a frequency domain to produce a spectrum, a power spectrum calculating unit configured to obtain powers of frequencies from the spectrum, a peak stationarity detecting unit configured to use peaks of the powers of frequencies in each frame to detect frequencies at which a stationary peak of the powers exists, a power stationarity detecting unit configured to use magnitudes of the powers of frequencies in each frame to detect frequencies at which the magnitudes of the powers are stationary, and a check unit configured to use the frequencies detected by the peak stationarity detecting unit and the frequencies detected by the power stationarity detecting unit to check whether there is a noise that has at least one of peak stationarity and power stationarity in the frequency domain.

Abstract: Provided is an audio encoding device capable of preventing audio quality degradation of a decoded signal. In the audio encoding device, a noise analysis unit (118) analyzes a noise characteristic of a higher range of an input spectrum. A filter coefficient decision unit (119) decides a filter coefficient in accordance with the noise characteristic information from the noise characteristic analysis unit (118). A filtering unit (113) includes a multi-tap pitch filter for filtering a first-layer decoded spectrum according to a filter state set by a filter state setting unit (112), a pitch coefficient outputted from a pitch coefficient setting unit (115), and a filter coefficient outputted from the filter coefficient decision unit (119), and calculates an estimated spectrum of the input spectrum. An optimal pitch coefficient can be decided by the process of a closed loop formed by the filter unit (113), a search unit (114), and the pitch coefficient setting unit (115).

Abstract: A high quality speech is reproduced with a small data amount in speech coding and decoding for performing compression coding and decoding of a speech signal to a digital signal. In speech coding method according to a code-excited linear prediction (CELP) speech coding, a noise level of a speech in a concerning coding period is evaluated by using a code or coding result of at least one of spectrum information, power information, and pitch information, and various excitation codebooks are used based on an evaluation result.

Abstract: In accordance with a specific implementation of the disclosure, a stream of audio frames is received and compressed using psycho-acoustical processing. The signal-to-mask ratio table generated by the psycho-acoustical algorithm is updated using only a portion of the received audio frames.

Abstract: Method for removing noise from a digital speech waveform, including receiving the digital speech waveform having the noise contained therein, segmenting the digital speech waveform into one or more frames, each frame having a clean portion and a noisy portion, extracting a feature component from each frame, creating an nonlinear speech distortion model from the feature components, creating a statistical noise model by making a Piecewise Linear Approximation (PLA) of the nonlinear speech distortion model, determining the clean portion of each frame using the statistical noise model, a log power spectra of each frame, and a model of a digital speech waveform recorded in a noise controlled environment, and constructing a clean digital speech waveform from each clean portion of each frame.

Abstract: A computer-readable medium recording a program allowing a computer to execute: setting a plurality of frames on a common time axis between a first waveform of an input to the audio processing and a second waveform of an output from the audio processing, detecting a voice frame and a noise frame in the first and second waveform, calculating a first and second spectrum from the first and second waveform, adjusting the level of the first or second spectrum of the noise frame, and setting the adjusted first and second spectrum of the noise frame as a third and fourth spectrum, calculating a distortion amount of the noise frame from the third and fourth spectrum, estimating a noise model spectrum from the first or second spectrum, and calculating a distortion amount of the voice frame from the first and second spectrum of the voice frame at the selected frequency.

Abstract: A speech recognition method comprising the steps of: storing multiple recognition models for a vocabulary set, each model distinguished from the other models in response to a Lombard characteristic, detecting at least one speaker utterance in a motor vehicle, selecting one of the multiple recognition models in response to a Lombard characteristic of the at least one speaker utterance, utilizing the selected recognition model to recognize the at least one speaker utterance; and providing a signal in response to the recognition.

Abstract: Preparing for the full-fledged aged society, measures to prevent senility are required. Senility is prevented by extracting signals of prescribed bands from a speech signal using a first bandpass filter section having a plurality of bandpass filters, extracting the envelopes of each frequency band signal using an envelope extraction section having envelope extractors, applying a noise source signal to a second bandpass filter section having a plurality of bandpass filters and extracting noise signals corresponding to the prescribed bands, multiplying the outputs from the first bandpass filter section and the second bandpass filter section in a multiplication section, summing up the outputs from the multiplication section in an addition section to produce a Noise-Vocoded Speech Sound signal, and presenting the Noise-Vocoded Speech Sound signal for listening.

Abstract: In a method (M) to detect a noise signal (PS1, PS2, PS3) in a digital audio signal (EAS), it is provided that the audio signal (EAS) is divided into successive signal sections (SAS), and the energy contents of successive signal sections (SAS) are determined, and the energy contents of a signal section (SAS) are evaluated in relation to an energy threshold (ET), and that the occurrence of at least one high-energy signal section having an energy content above the energy threshold (ET), and the occurrence of at least one signal section (SAS) preceding the at least one high-energy signal section and having an energy content below the energy threshold (ET), and the occurrence of at least one signal section (SAS) following the at least one high-energy signal section and having an energy content below the energy threshold (ET) are detected, and that a quantity of signal sections (SAS) that precede the at least one high-energy signal section and a quantity of high-energy signal sections and a quantity of signal secti

Abstract: Provided is a code conversion device that is capable of converting codes even if an input code sequence is invalid, and is able to reduce the amount of processing. When a first code sequence is input, the code conversion device generates a decoded signal by decoding the codes of normal frames of the first code sequence at Step S1, stores and holds the decoded signal at Step S2, generates a signal corresponding to an invalid frame by interpolation with the decoded signal that is stored and held, at Step S3. Subsequently, the code conversion device generates codes corresponding to the invalid frame by encoding the generated signal at Step S4, and makes the normal frames of the first code sequence without conversion be the frames of the second code sequence while making the generated codes be the frame of the second code sequence, in place of the codes of the invalid frame, at Step S5.

Abstract: A noise robustness method operates jointly in a signal domain and a model domain. For example, energy is added in the signal domain for frequency bands where an actual noise level of an incoming signal is lower than a noise level used to train models, thus obtaining a compensated signal. Also, energy is added in the model domain for frequency bands where noise level of the incoming signal or the compensated signal is higher than the noise level used to train the models. Moreover, energy is never removed, thereby avoiding problems of higher sensitivity of energy removal to estimation errors.

Abstract: A method and apparatus identify a clean speech signal from a noisy speech signal. To do this, a clean speech value and a noise value are estimated from the noisy speech signal. The clean speech value and the noise value are then used to define a gain on a filter. The noisy speech signal is applied to the filter to produce the clean speech signal. Under some embodiments, the noise value and the clean speech value are used in both the numerator and the denominator of the filter gain, with the numerator being guaranteed to be positive.

Abstract: A spectrum of a set of samples from a data stream of sampled data is estimated until a targeted signal to noise ratio is achieved.

Abstract: Provided is a method for canceling background noise of a sound source other than a target direction sound source in order to realize highly accurate speech recognition, and a system using the same. In terms of directional characteristics of a microphone array, due to a capability of approximating a power distribution of each angle of each of possible various sound source directions by use of a sum of coefficient multiples of a base form angle power distribution of a target sound source measured beforehand by base form angle by using a base form sound, and power distribution of a non-directional background sound by base form, only a component of the target sound source direction is extracted at a noise suppression part. In addition, when the target sound source direction is unknown, at a sound source localization part, a distribution for minimizing the approximate residual is selected from base form angle power distributions of various sound source directions to assume a target sound source direction.

Abstract: An enhancement system improves the perceptual quality of a processed speech. The system includes a delay unit that delays a signal received through a discrete input. A spectral modifier linked to the delay unit is programmed to substantially flatten the spectral character of a background noise. An adaptive filter linked to the spectral modifier adapts filter characteristics to match a response of a non-delayed signal. A programmable filter is linked to the delay unit. The programmable filter has a transfer function functionally related to a transfer function of the adaptive filter.

Abstract: An apparatus and method for detecting a voice activity period. The apparatus for detecting a voice activity period includes a domain conversion module that converts an input signal into a frequency domain signal in the unit of a frame obtained by dividing the input signal at predetermined intervals, a subtracted-spectrum-generation module that generates a spectral subtraction signal which is obtained by subtracting a predetermined noise spectrum from the converted frequency domain signal, a modeling module that applies the spectral subtraction signal to a predetermined probability distribution model, and a speech-detection module that determines whether a speech signal is present in a current frame through a probability distribution calculated by the modeling module.

Abstract: An audio signal reduction device that generates a gap period in accordance with a noise generation period of noise included in an input audio signal. Noise is removed from the audio signal, and the level envelope of the audio signal is continuously detected. A coefficient for the level envelope in the gap period is generated in accordance with the level envelope detection and is used to modulate an interpolated signal. The noise-removed audio signal and the modulated interpolated signal are mixed; and the mixed signal is output in a period corresponding to the gap period, while the audio signal is output, as is, not in the gap period.

Abstract: Input speech is coded in an encoder (11), the coded speech is decoded in a decoder (12), compensatory speech which compensates the speech of the current frame is generated in a compensatory speech generating part (20) by using past decoded speech, the quality of the compensatory speech is evaluated by using the input speech and the compensatory speech and a duplication level is generated the value of which increases incrementally with decreasing speech quality evaluation value in a speech quality evaluating part (40), and as many identical packets as the number specified by the duplication level is generated for the coded speech in a packet generating part (15), and the packets are transmitted, thereby reducing the possibility that packet loss will occur at the receiving end.

Abstract: A sound processing system including a user headset for use in tactical military operations provides integrated sound and speech analysis including sound filtering and amplification, sound analysis and speech recognition for analyzing speech and non-speech sounds and taking programmed actions based on the analysis, recognizing language of speech for purposes of one-way and two-way voice translation, word spotting to detect and identify elements of conversation, and non-speech recognition and identification. The headset includes housings with form factors for insulating a user's ear from direct exposure to ambient sounds with at least one microphone for receiving sound around the user, and a microphone for receiving user speech. The user headset can further include interconnections for connecting the headset with out systems outside of the headset, including target designation systems, communication networks, and radio transmitters.

Abstract: Disclosed are an apparatus and a method for voice processing in a mobile communication terminal. A plurality of microphones are used to remove environmental noise at the time of voice communication, so that it is possible to perform high-quality voice communication and video telephony. Moreover, it is possible to perform voice recording even when a user does not open a mobile communication terminal. Furthermore, when voice is recorded or sound is recorded during moving image photographing, a plurality of microphones are effectively utilized to achieve good-quality recording and to perform recording conveniently even when the folder or the slider of the mobile communication terminal is closed. Therefore, it is possible to provide improved convenience in using the mobile communication terminal.

Abstract: The present invention is a system and method for packetizing actual noise signals, typically background noise, received by an access gateway from a speaking party and transmitting these packetized noise signals via a network to an egress gateway. The egress gateway converts the packetized noise signal into noise signals suitable for output and transmits the output noise signals to a listening party. When the access gateway detects that no voice signal is being received and only a noise signal is being received for a predetermined period of time, the access gateway instructs the egress network to continually transmit output noise signals to the listening party and ceases to transmit packetized noise signals to the egress gateway.

Abstract: First encoded voice bits are transcoded into second encoded voice bits by dividing the first encoded voice bits into one or more received frames, with each received frame containing multiple ones of the first encoded voice bits. First parameter bits for at least one of the received frames are generated by applying error control decoding to one or more of the encoded voice bits contained in the received frame, speech parameters are computed from the first parameter bits, and the speech parameters are quantized to produce second parameter bits. Finally, a transmission frame is formed by applying error control encoding to one or more of the second parameter bits, and the transmission frame is included in the second encoded voice bits.

Abstract: Provided are an apparatus and method for estimating noise and a noise reduction apparatus employing the same. The noise estimation apparatus estimates noise by blocking audio signals from a direction of a target sound source from received audio signals, and compensating for distortions from directivity gains of a target sound blocker blocking the audio signals from the target sound source.

Abstract: A noise reduction device is configured by use of: means for calculating a predetermined constant, and a predetermined reference signal R?(T) in the frequency domain, respectively by use of adaptive coefficients W?(m), and for thereby obtaining estimated values N? and Q?(T) respectively of stationary noise components, and non-stationary noise components corresponding to the reference signal, which are included in a predetermined observed signal X?(T) in the frequency domain; means and for applying a noise reduction process to the observed signal on the basis of each of the estimated values, and for updating each of the adaptive coefficients on the basis of a result of the process; and an adaptive learning means and for repeating the obtaining of the estimated values and the updating of the adaptive coefficients, and for thereby learning each of the adaptive coefficients.