Abstract: There is provided a speech post-processor for enhancing a speech signal divided into a plurality of sub-bands in frequency domain. The speech post-processor comprises an envelope modification factor generator configured to use frequency domain coefficients representative of an envelope derived from the plurality of sub-bands to generate an envelope modification factor for the envelope derived from the plurality of sub-bands, where the envelope modification factor is generated using FAC=?ENV/Max+(1??), where FAC is the envelope modification factor, ENV is the envelope, Max is the maximum envelope, and ? is a value between 0 and 1, where ? is a different constant value for each speech coding rate. The speech post-processor further comprises an envelope modifier configured to modify the envelope derived from the plurality of sub-bands by the envelope modification factor corresponding to each of the plurality of sub-bands.

Abstract: Techniques and tools related to delayed or lost coded audio information are described. For example, a concealment technique for one or more missing frames is selected based on one or more factors that include a classification of each of one or more available frames near the one or more missing frames. As another example, information from a concealment signal is used to produce substitute information that is relied on in decoding a subsequent frame. As yet another example, a data structure having nodes corresponding to received packet delays is used to determine a desired decoder packet delay value.

Abstract: A decoder for code excited LP encoded frames with both adaptive and fixed codebooks; erased frame concealment uses repetitive excitation plus a smoothing of pitch gain in the next good frame, plus multilevel voicing classification with multiple thresholds of correlations determining linear interpolated adaptive and fixed codebook excitation contributions.

Abstract: A post filter and a decoder enabling improvement of the sound quality of a decoded signal even when the sound quality of the decoded signal is different from the bands are disclosed. A frequency converting section determines a decoded spectrum. A power spectrum computing section computes the power spectrum from the decoded spectrum. A correction band determining section determines the band in which the power spectrum is corrected according to layer information. A power spectrum correcting section corrects the power spectrum in the corrected band in such a way that the variation along the frequency axis is suppressed. An inverse converting section subjects the corrected power spectrum to inverse conversion to determine an autocorrelation function. An LPC analyzing section determines an LPC coefficient of the determined autocorrelation function.

Abstract: Square sum calculator 603 calculates a square sum of evolution in smoothed quantized LSP parameter for each order. A first dynamic parameter is thereby obtained. Square sum calculator 605 calculates a square sum using a square value of each order. The square sum is a second dynamic parameter. Maximum value calculator 606 selects a maximum value from among square values for each order. The maximum value is a third dynamic parameter. The first to third dynamic parameters are output to mode determiner 607, which determines a speech mode by judging the parameters with respective thresholds to output mode information.

Abstract: The present invention provides a method for recovering target speech based on shapes of amplitude distributions of split spectra obtained by use of blind signal separation.

Abstract: A method for providing recognition error correction information, the method includes: obtaining metadata associated with a capture of a media item; and generating recognition error correction information in response to the metadata. The recognition error correction information is to be used in a recognition process selected out of a list consisting of an automatic speech recognition process and an optical characters recognition process.

Abstract: A method and apparatus are provided for reducing noise in a signal. Under one aspect of the invention, a correction vector is selected based on a noisy feature vector that represents a noisy signal. The selected correction vector incorporates dynamic aspects of pattern signals. The selected correction vector is then added to the noisy feature vector to produce a cleaned feature vector. In other aspects of the invention, a noise value is produced from an estimate of the noise in a noisy signal. The noise value is subtracted from a value representing a portion of the noisy signal to produce a noise-normalized value. The noise-normalized value is used to select a correction value that is added to the noise-normalized value to produce a cleaned noise-normalized value. The noise value is then added to the cleaned noise-normalized value to produce a cleaned value representing a portion of a cleaned signal.

Abstract: The invention relates to a network element (1) and a method for enhancing the quality of digitised analogue signals transmitted in parameterised coded form via a digital network. In order to enable an enhancement of the quality of the digitised analogue signals on network side, the network element comprises means (20, 21) for extracting signals from and insert signals into the network, first processing means (24) for processing the extracted parameters in the parameter domain with functions suitable to enhance the quality of the digitised analogue signals and second processing means (26) for processing the extracted parameters in the linear domain with functions suitable to enhance the quality of the digitised analogue signals. Moreover included analysing and selecting means (23, 27) determine the expected enhancement of quality in the different processing domains and cause a corresponding insertion of processed signals back into the network. The proposed method comprises corresponding steps.

Abstract: A system and method to generate a trigger signal based on a real-time adaptive threshold. The system may include a microphone to receive an audio signal, a device to generate a trigger signal based on a real-time adaptive threshold coupled to the microphone to form an adaptive threshold and generate a trigger signal if a magnitude of the audio signal is greater than a magnitude of the adaptive threshold. The system may also include a waveform capture module coupled to the microphone to receive the audio signal and convert the audio signal into a series of waveform packets and a waveform analysis processor to extract characteristics from the waveform packets.

Abstract: Method for recovering target speech by extracting signal components falling in a speech segment, which is determined based on separated signals obtained through the Independent Component Analysis, thereby minimizing the residual noise in the recovered target speech.

Abstract: An entropy encoder includes an apparatus for producing a data stream which comprises two reference points, of code words of variable lengths, the apparatus comprising a first device for writing at least a part of a code word into the data stream in a first direction of writing, starting from a first reference point, and a second device for writing at least a part of a code word into the data stream in a second direction of writing, which is opposite to the first direction of writing, starting from the other reference point. In particular, when a raster having a plurality of segments is used to write the code words of variable lengths into the data stream, the number of the code words which can be written starting at raster points is doubled, in the best case, such that the data stream of code words of variable lengths is robust toward a propagation of sequence errors.

Abstract: A system and method are provided that reduce noise in speech signals. The system and method decompose a noisy speech signal into a harmonic component and a residual component. The harmonic component and residual component are then combined as a sum to form a noise-reduced value. In some embodiments, the sum is a weighted sum where the harmonic component is multiplied by a scaling factor. In some embodiments, the noise-reduced value is used in speech recognition.

Abstract: A filter controller processes a decoded speech (DS) signal. The DS signal has a spectral envelope including a first plurality of formant peaks having different respective amplitudes. The controller produces, from the DS signal, a spectrally-flattened DS signal that is a time-domain signal. The spectrally-flattened time-domain DS signal has a spectral envelope including a second plurality of formant peaks. Each of the second plurality of formant peaks approximately coincides in frequency with a respective one of the first plurality of formant peaks. Also, the second plurality of formant peaks have approximately equal respective amplitudes. Next, the controller derives, from the spectrally-flattened time-domain DS signal, a set of filter coefficients representative of a filter response that is to be used to filter the DS signal.

Abstract: An apparatus for processing a speech signal includes a receiver, a speech signal decoder, a speech rate conversion information detector, and a speech rate converting processor. The receiver receives multiplexed signal of information concerning controls and programs, including speech packets through a transmission line. The decoder decodes the speech signal of packets out of the received signals. The detector detects speech rate conversion execution information in the received signals. The processor subjects the decoded speech signal to a speech rate conversion process if the speech rate conversion execution information indicates that the speech signal has not been subjected to the speech rate conversion process on the transmitting end, and which does not subject the decoded speech signal to the speech rate conversion process if the speech rate conversion execution information indicates that the speech signal has been subjected to the speech rate conversion process on the transmitting end.

Abstract: A speech signal transmission apparatus multiplexes, packetizes, and sends first coded information coded in a normal state and second coded information used for improving the quality of decoded speech when a frame loss occurs. A first error calculating section calculates a first error signal between a target signal and a synthesized signal generated by an adaptive codebook, and a second error calculating section calculates a second error signal between the target signal and a synthesized signal generated by a fixed codebook. An error signal ratio calculating section calculates the ratio of the first error signal to the second error signal. A speech frame classifying section classifies a speech frame according to the magnitude of the ratio, and a decision section decides whether or not to multiplex the second coded information based on the classification result.

Abstract: The delay time and listening quality of a system under test are measured from a signal received therefrom, then the measured delay time and listening quality are transformed to a delay-related degradation and a listening quality degradation on the same quality measure, then the quantity of interaction between the delay-related degradation and the listening quality degradation is calculated, and the delay-related degradation, the listening quality degradation and the quantity of interaction are added together to obtain an overall degradation. The overall degradation is transformed to a subjective evaluation value to estimate the overall speech quality.

Abstract: A system is described that performs periodic waveform extrapolation based frame erasure concealment (FEC) to generate frames of an output speech signal corresponding to erased frames of encoded bit-stream in a manner reduces buzzy and tonal artifacts in the output speech signal. An embodiment of the invention uses a multiple of a pitch period associated with previously-decoded speech to perform periodic waveform extrapolation for consecutively-erased frames in a frame erasure beyond the first erased frame. An embodiment of the invention also attenuates the extrapolated signal after a threshold number of erased frames so as to reduce the FEC output signal to zero, wherein the threshold number of erased frames is dependent at least in part on the pitch period associated with the previously-decoded speech.

Abstract: An audio data reproduction apparatus, integrated circuit device, and computer program product configured to implement a method for reproducing audio data that includes receiving an audio stream, decoding the audio stream, processing the audio stream or a signal produced by the decoding, selecting the signal produced by the decoding or a signal produced by the processing, and outputting the audio stream, the signal produced by the decoding, or the signal produced by the processing.

Abstract: To alleviate problems of signal aliasing and to reduce complexity, Linear Predictive Coefficients (LPCS) are calculated from samples of audio signals and Line Spectral Frequency (LSF) vectors are extracted from the LPCs with a rate higher than a desired vector rate, the LSF vectors comprising values of different LSF parameters. Next, an LSF track is formed for at least one of the LSF parameters. At least one of the formed LSF tracks is then low pass filtered. Finally, decimated LSF vectors are reconstructed from the low pass filtered LSF tracks, the decimated number corresponding to the desired vector rate. The invention equally relates to a corresponding computer program, to corresponding devices and to a corresponding communication network.

Abstract: An audio system for processing two channels of audio input to provide more than two output channels. The input may be conventional stereo material or compressed audio signal data. The audio processing includes separating the input signals into frequency bands and processing the frequency bands according to processes which may differ from band to band. The audio processing includes no processing of L?R signals.

Abstract: A testing arrangement provided for speech recognition systems in vehicles. Preferably included are a “mobile client” secured in the vehicle and driven around at a desired speed, an audio system and speaker which plays back a set of prerecorded utterances stored digitally in a computer arrangement such that the speech of a human being is simulated, transmission of the speech signal to a server, followed by speech recognition and signal-to-noise ratio (SNR) computation. Here, the acceptability of the vehicular speech recognition system is preferably determined via comparison with pre-specified standards of recognition accuracy and SNR values.

Abstract: A method for adaptive long-term filtering of an audio signal, such as a decoded speech signal. The method includes measuring a smoothed periodicity of an audio signal segment, such as an audio frame, wherein the smoothed periodicity is measured by low-pass filtering an instantaneous periodicity of the audio signal segment. The periodicity of the audio signal segment is then increased in a manner that depends upon whether the smoothed periodicity is less than a predetermined threshold. By utilizing a smoothed periodicity measurement in this fashion, more accurate control of the post-filter is provided as compared to conventional solutions. Additionally, the method includes deriving filters by interpolating between filter responses of adjacent audio signal segments to minimize distortion at segment boundaries.

Abstract: A multi-channel audio decoder provides a reduced complexity processing to reconstruct multi-channel audio from an encoded bitstream in which the multi-channel audio is represented as a coded subset of the channels along with a complex channel correlation matrix parameterization. The decoder translates the complex channel correlation matrix parameterization to a real transform that satisfies the magnitude of the complex channel correlation matrix. The multi-channel audio is derived from the coded subset of channels via channel extension processing using a real value effect signal and real number scaling.

Abstract: A method for improving the quality of a speech signal extracted from a noisy acoustic environment is provided. In one approach, a signal separation process is associated with a voice activity detector. The voice activity detector is a two-channel detector, which enables a particularly robust and accurate detection of voice activity. When speech is detected, the voice activity detector generates a control signal. The control signal is used to activate, adjust, or control signal separation processes or post-processing operations to improve the quality of the resulting speech signal. In another approach, a signal separation process is provided as a learning stage and an output stage. The learning stage aggressively adjusts to current acoustic conditions, and passes coefficients to the output stage. The output stage adapts more slowly, and generates a speech-content signal and a noise dominant signal.

Abstract: Extracting features from signals for use in classification, retrieval, or identification of data represented by those signals uses a “Distortion Discriminant Analysis” (DDA) of a set of training signals to define parameters of a signal feature extractor. The signal feature extractor takes signals having one or more dimensions with a temporal or spatial structure, applies an oriented principal component analysis (OPCA) to limited regions of the signal, aggregates the output of multiple OPCAs that are spatially or temporally adjacent, and applies OPCA to the aggregate. The steps of aggregating adjacent OPCA outputs and applying OPCA to the aggregated values are performed one or more times for extracting low-dimensional noise-robust features from signals, including audio signals, images, video data, or any other time or frequency domain signal. Such extracted features are useful for many tasks, including automatic authentication or identification of particular signals, or particular elements within such signals.

Abstract: A method according to the invention separates multiple audio signals recorded as a mixed signal via a single channel. The mixed signal is A/D converted and sampled. A sliding window is applied to the samples to obtain frames. The logarithms of the power spectra of the frames are determined. From the spectra, the a posteriori probabilities of pairs of spectra are determined. The probabilities are used to obtain Fourier spectra for each individual signal in each frame. The invention provides a minimum-mean-squared error metho or a soft mask method for making this determination. The Fourier spectra are inverted to obtain corresponding signals, which are concatenated to recover the individual signals.

Abstract: A common problem in audio processing is that a useful signal is disturbed by one or more sinusoidal noises that should be suppressed. One embodiment of the invention provides a method of canceling a sinusoidal disturbance of unknown frequency in a disturbed useful signal. The method comprises the steps of estimating parameters of the sinusoidal disturbance including amplitude, phase and frequency; generating a reference signal on the basis of the estimated parameters; and subtracting the reference signal from the disturbed useful signal. According to one embodiment of the present invention, the estimation is performed by an Extended Kalman filter.

Abstract: An encoder transforms at least a portion of a signal, counts the resulting transform coefficients having a zero value, and encodes the signal with the zero count. A decoder decodes the signal in order to recover the zero count. The decoder may also determine its own zero count of the signal as received and may compare the zero count that it determines to the recovered zero count. The decoder may be arranged to detect compression/decompression based upon results from the comparison, and/or the decoder may be arranged to prevent use of a device based upon results from the comparison.

Abstract: A method and system use an alternative sensor signal received from a sensor other than an air conduction microphone to estimate a clean speech value. The estimation uses either the alternative sensor signal alone, or in conjunction with the air conduction microphone signal. The clean speech value is estimated without using a model trained from noisy training data collected from an air conduction microphone. Under one embodiment, correction vectors are added to a vector formed from the alternative sensor signal in order to form a filter, which is applied to the air conductive microphone signal to produce the clean speech estimate. In other embodiments, the pitch of a speech signal is determined from the alternative sensor signal and is used to decompose an air conduction microphone signal. The decomposed signal is then used to determine a clean signal estimate.

Abstract: In a method of extracting voice components free of noise components from voice signals input through a single microphone, a signal-decomposing unit extracts independent signal components from the voice signals input through a single microphone by using a plurality of filters that permit the passage of signal components of different frequency bands. A signal-synthesizing unit synthesizes the signal components according to a first rule to form a first synthesized signal, and synthesizes the signal components according to a second rule to form a second synthesized signal. The first and second rules are so determined that a difference becomes a maximum between the probability density function of the first synthesized signal and the probability density function of the second synthesized signal. An output selection unit selectively produces a synthesized signal having a large difference from the Gaussian distribution between the synthesized signals.

Abstract: Minimizing the effects of the requisite AGWN packets on transmission channel utilization without diminishing any of the aesthetic quality of the AGWN white noise on the voice or audio communication.

Abstract: In various embodiments of the present invention, a noisy signal denoiser is tuned and optimized by selecting denoiser parameters that provide relatively highly compressible denoiser output. When the original signal can be compared to the output of a denoiser, the denoiser can be accurately tuned and adjusted in order to produce a denoised signal that resembles as closely as possible the clear signal originally transmitted through a noise-introducing channel. However, when the clear signal is not available, as in many communications applications, other methods are needed. By adjusting the parameters to provide a denoised signal that is globally or locally maximally compressible, the denoiser can be optimized despite inaccessibility of the original, clear signal.

Abstract: An entropy encoder includes an apparatus for producing a data stream which comprises two reference points, of code words of variable lengths, the apparatus comprising a first device for writing at least a part of a code word into the data stream in a first direction of writing, starting from a first reference point, and a second device for writing at least a part of a code word into the data stream in a second direction of writing, which is opposite to the first direction of writing, starting from the other reference point. In particular, when a raster having a plurality of segments is used to write the code words of variable lengths into the data stream, the number of the code words which can be written starting at raster points is doubled, in the best case, such that the data stream of code words of variable lengths is robust toward a propagation of sequence errors.

Abstract: At an audio source, pause information is added to audio data, the combination of which is subsequently packetized. The resulting packets are transmitted to an audio destination via a network in which different packets may be subjected to varying levels of delay. At the audio destination, the pause information may be used to insert pauses at appropriate times to accommodate the occurrence of delays in packet delivery. In one embodiment, pauses are inserted based on a hierarchy of pause types. During pauses, audio filler information may be injected. In this manner, the effects of variable network delays upon reconstructed audio may be mitigated.

Abstract: Digital audio data with error detection bits added thereto is inputted to an error detecting and correcting device (4). The correcting device (4) corrects an error when the error is detected in the digital audio data. The digital audio data outputted from the error detecting and correcting device (4) is inputted to an impulse noise suppressing circuit (6). The suppressing circuit (6) operates for a predetermined time period when the correcting device (4) detects an error.

Abstract: In a speech signal decoding method, information containing at least a sound source signal, gain, and filter coefficients is decoded from a received bit stream. Voiced speech and unvoiced speech of a speech signal are identified using the decoded information. Smoothing processing based on the decoded information is performed for at least either one of the decoded gain and decoded filter coefficients in the unvoiced speech. The speech signal is decoded by driving a filter having the decoded filter coefficients by an excitation signal obtained by multiplying the decoded sound source signal by the decoded gain using the result of the smoothing processing. A speech signal decoding apparatus is also disclosed.

Abstract: A method and apparatus for assessing the perceptual quality of stereo speech signals transmitted via a telecommunications network and recorded acoustically from an acoustic terminal device in which a mono reference signal comprising a single channel is aligned with a degraded stereo signal comprising a left and a right channel; a delay between each channel of said degraded signal and said reference signal is estimated; a noise masking indicator in dependence upon said estimated delays is generated; the level of the stereo signals is adjusted in dependence upon said noise masking indicator; a set of perceptually relevant parameters for each of said reference and degraded signals are generated; the perceptually relevant parameters of the reference signal with the perceptually relevant parameters of the degraded signal to generate a disturbance profile are compared; and a speech quality prediction is generated in dependence upon said disturbance profile.

Abstract: A system and method of concealing bit errors in a signal are provided. An exemplary method detects bit errors in an input signal having at least a current signal segment and a previous signal segment. The previous signal segment has a log-gain value qlg(m?1) and immediately precedes the current signal segment. The method comprises estimating a level lvl(m?1) of the input signal and determining a log-gain value qlg(m) of the current signal segment within the input signal. The method also comprises determining a difference between the gain value of the current signal segment and the previous signal segment and determining whether the difference exceeds a threshold. wherein the threshold is adaptive to the input signal level.

Abstract: A control system receiving an input signal, comprising speech with ambient noise, determines the ambient noise power spectrum, retrieves control information corresponding to the closest-matching ambient noise power spectrum, and outputs the input signal along with display of a predetermined effect or image corresponding to the ambient noise power spectrum.

Abstract: The invention relates to method of processing a speech frame in radio system, a radio system, a mobile station in radio system, and a network of radio system. In the method a speech frame having propagated over a radio path is channel-decoded. If on the basis of the channel-decoding the speech frame is free of defects, it is inferred from the value of at least one speech parameter in the channel-decoded speech frame whether the speech frame contains speech that is decodable by means of a speech decoder; and if, according to the inference, the speech frame does contain speech that is decodable by means of a speech decoder, the speech frame is decoded by means of a speech decoder; and if, according to the inference, the speech frame does not contain speech that would be decodable by means of the speech decoder, the speech frame is not decoded.

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 method and apparatus for robust speech activity detection is disclosed. The method may include calculating autocorrelations by filtering input signals using order statistic filtering, averaging the autocorrelations over a time period, obtaining a voiced speech feature from the averaged autocorrelations, classifying the input signal as one of speech and non-speech based on the obtained voiced speech feature, and outputting only the classified speech signals or the input signals along with the speech/non-speech classification information, to an automated speech recognizer.

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 pitch adaptive circuit (200) includes an equalizer control circuit (206) that evaluates the pitch of the speech signals that are being processed and depending on the pitch information, the equalizer control circuit (206) selects an equalizer (208, 210) to shape the decoded speech signals. By selecting the best equalizer (208 or 210) to use based on the pitch information, improvements in audio quality are provided automatically without user intervention.

Abstract: A method of reducing noise by cascading a plurality of noise reduction algorithms is provided. A sequence of noise reduction algorithms are applied to the noisy signal. The noise reduction algorithms are cascaded together, with the final noise reduction algorithm in the sequence providing the system output signal. The sequence of noise reduction algorithms includes a plurality of noise reduction algorithms that are sufficiently different from each other such that resulting distortions and artifacts are sufficiently different to result in reduced human perception of the artifact and distortion levels in the system output signal.

Abstract: A frame erasure concealment device and method that is based on reestimating gain parameters for a code excited linear prediction (CELP) coder is disclosed. During operation, when a frame in a stream of received data is detected as being erased, the coding parameters, especially an adaptive codebook gain gp and a fixed codebook gain gc, of the erased and subsequent frames can be reestimated by a gain matching procedure. By using this technique with the IS-641 speech coder, it has been found that the present invention improves frame erasure concealment device and method improve the speech quality under various channel conditions, compared with a conventional extrapolation-based concealment algorithm.

Abstract: An approach for efficiently reducing background noise from speech signal in real-time applications is presented. A noisy input speech signal is processed through an inverse filter when the spectrum tilt of the input signal is not that of a pure background noise model the noisy input signal is also filtered in order to reduce the spectrum valley areas of the noisy input signal when the background noise is present.

Abstract: An automated speech recognition filter is disclosed. The automated speech recognition filter device provides a speech signal to an automated speech platform that approximates an original speech signal as spoken into a transceiver by a user. In providing the speech signal, the automated speech recognition filter determines various models representative of a cumulative signal degradation of the original speech signal from various devices along a transmission signal path and a reception signal path between the transceiver and a device housing the filter. The automated speech platform can thereby provide an audio signal corresponding to a context of the original speech signal.

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.