Abstract: Embodiments of the present disclosure are directed to techniques for adjusting the amplitude of a digital audio signal in the frequency domain to control the perceived loudness of the audio signal at a desired level. In one embodiment, a method first adjusts the audio signal to a desired loudness level by applying an adaptive wideband gain and thereafter a multi-band compression is applied to further reduce a dynamic range of the audio signal, and noise analysis and temporal masking operations are also performed to provide a pleasant sound for a listener or listeners.

Abstract: Embodiments reduce the complexity of speaker dependent speech recognition systems and methods by representing the code phrase (i.e., the word or words to be recognized) using a single Gaussian Mixture Model (GMM) which is adapted from a Universal Background Model (UBM). Only the parameters of the GMM need to be stored. Further reduction in computation is achieved by only checking the GMM component that is relevant to the keyword template. In this scheme, keyword template is represented by a sequence of the index of best performing component of the GMM of the keyword model. Only one template is saved by combining the registration template using Longest Common Sequence algorithm. The quality of the word model is continuously updated by performing expectation maximization iteration using the test word which is accepted as keyword model.

Abstract: Methods and systems of text independent speaker recognition provide a complexity comparable to text dependent speaker recognition system. These methods and systems exploit the fact that speech is a quasi-stationary signal and simplify the recognition process based on this theory. The speaker modeling allows a speaker profile to be updated progressively with new speech samples that are acquired during usage over time by the speaker.

Abstract: A method of estimating a steering vector of a sensor array of M sensors according to one embodiment of the present disclosure includes estimating a steering vector of a noise source located at an angle ? degrees from a look direction of the array using a least squares estimate of the gains of the sensors in the array, defining a steering vector of a desired sound source in the look direction of the array, and estimating the steering vector by performing element-by-element multiplication of the estimated noise vector and the complex conjugate of steering vector of the desired sound source. The sensors may be microphones.

Abstract: The present invention is a system and method for digital watermarking, which discloses a system for digital watermarking, to add a watermark to an audio signal generated by a signal source. The system comprises: a spectrum modulator configured to perform spectrum modulation to a watermark bit and a pseudo noise signal to be embedded into the audio signal to generate a modulated signal; a distortion controller coupled to the signal source and the spectrum modulator and configured to shape the modulated signal based on the audio signal, so as to generate a shaped signal satisfying a predetermined distortion constraint; and an interference compensator coupled to the signal source and the distortion controller and configured to generate a compensation signal based on the audio signal, the pseudo noise signal, and the shaped signal, wherein the compensation signal is for compensating for interference to watermark decoding caused by the audio signal.

Abstract: Embodiments of the present disclosure are directed to techniques for adjusting the amplitude of a digital audio signal in the frequency domain to control the perceived loudness of the audio signal at a desired level. In one embodiment, a method first adjusts the audio signal to a desired loudness level by applying an adaptive wideband gain and thereafter a multi-band compression is applied to further reduce a dynamic range of the audio signal, and noise analysis and temporal masking operations are also performed to provide a pleasant sound for a listener or listeners.

Abstract: A method of estimating a steering vector of a sensor array of M sensors according to one embodiment of the present disclosure includes estimating a steering vector of a noise source located at an angle 0 degrees from a look direction of the array using a least squares estimate of the gains of the sensors in the array, defining a steering vector of a desired sound source in the look direction of the array, and estimating the steering vector by performing element-by-element multiplication of the estimated noise vector and the complex conjugate of steering vector of the desired sound source. The sensors may be microphones.

Abstract: Methods and apparatus for self-calibration of small-microphone arrays are described. In one embodiment, self-calibration is based upon a mathematical approximation for which a detected response by one microphone should approximately equal a combined response from plural microphones in the array. In a second embodiment, self-calibration is based upon matching gains in each of a plurality of Bark frequency bands, and applying the matched gains to frequency domain microphone signals such that the magnitude response of all the microphones in the array approximates an average magnitude response for the array. The methods and apparatus may be implemented in hearing aids or small audio devices and used to mitigate adverse aging and mechanical effects on acoustic performance of small-microphone arrays in these systems.

Abstract: The present invention is a system and method for digital watermarking, which discloses a system for digital watermarking, to add a watermark to an audio signal generated by a signal source. The system comprises: a spectrum modulator configured to perform spectrum modulation to a watermark bit and a pseudo noise signal to be embedded into the audio signal to generate a modulated signal; a distortion controller coupled to the signal source and the spectrum modulator and configured to shape the modulated signal based on the audio signal, so as to generate a shaped signal satisfying a predetermined distortion constraint; and an interference compensator coupled to the signal source and the distortion controller and configured to generate a compensation signal based on the audio signal, the pseudo noise signal, and the shaped signal, wherein the compensation signal is for compensating for interference to watermark decoding caused by the audio signal.

Abstract: The perception of 3D sound positioning can be achieved using a 2D arrangement of speakers positioned around the listener. The disclosed techniques can enable listeners to perceive sounds as coming from above and/or below them, without the need for positioning speakers above and/or below the listener. In some embodiments, elevation information can be included in the X and Y horizontal components of the 2D ambisonics encoding. The X and Y components can be decoded using 2D ambisonics decoding. Suitable filtering may be performed on the decoded sound information to enhance the listener's perception of the elevation information encoded in the X and Y components.

Abstract: A time-domain system and method of modifying the time scale of digital audio signals includes a pre-processor. The pre-processor forms a synthesized signal for processing with minimum computation and that has optional features to give preference to certain audio channels and/or frequency bands, a mechanism of adaptively characterizing the temporal features of the synthesized signal by its normalized power and zero-crossing count, and a mechanism of identifying a segment of the synthesized signal where the time scale can be modified without introducing artifacts or losing content.

Abstract: A system and method for generating virtual microphone signals having a particular number and configuration for channel playback from an intermediate set of signals that were recorded in an initial format that is different from the channel playback format. In one embodiment, an initial set of intermediate are Bark-banded such that each intermediate signal may lead to a corresponding power spectral density (PSD) signal representative of the initial intermediate signal. Further, one may generate cross-correlations signals for each pair of intermediate signals. Next, from the PSDs and cross correlations, one may more efficiently calculate corresponding channel signals to be used for playback on respective channel speakers. Thus, the PSDs of each channel signal may be generated at chosen angles (as well as other design factors). Further, each channel signal may also be further modified with a corresponding cancellation signal that further enhances the resultant signal in each channel.

Abstract: In an embodiment, an apparatus includes a determiner, converter, adapter, and modifier. The determiner is configured to generate a representation of a difference between a first frequency at which a first signal is sampled and a second frequency at which a second signal is sampled, and the converter is configured to generate a second sample of the first signal at a second time in response to the representation and a first sample of the first signal at a first time. The adapter is configured to generate a sample of a modifier signal in response to the second sample of the first signal, and the modifier is configured to generate a modified sample of the second signal in response to a sample of the second signal and the sample of the modifier signal. For example, such an apparatus may be able to reduce the magnitude of an echo signal in a system having an audio pickup (e.g., a microphone) near an audio output (e.g., a speaker).

Abstract: Embodiments reduce the complexity of speaker dependent speech recognition systems and methods by representing the code word (i.e., the word to be recognized) using a single Gaussian Mixture Model (GMM) which is adapted from a Universal Background Model (UBM). Only the parameters of the GMM need to be stored. Further reduction in computation is achieved by only checking the GMM component that is relevant to the keyword template. In this scheme, keyword template is represented by a sequence of the index of best performing component of the GMM of the keyword model. Only one template is saved by combining the registration template using Longest Common Sequence algorithm. The quality of the word model is continuously updated by performing expectation maximization iteration using the test word which is accepted as keyword model.

Abstract: In an embodiment, an apparatus includes a determiner, converter, adapter, and modifier. The determiner is configured to generate a representation of a difference between a first frequency at which a first signal is sampled and a second frequency at which a second signal is sampled, and the converter is configured to generate a second sample of the first signal at a second time in response to the representation and a first sample of the first signal at a first time. The adapter is configured to generate a sample of a modifier signal in response to the second sample of the first signal, and the modifier is configured to generate a modified sample of the second signal in response to a sample of the second signal and the sample of the modifier signal. For example, such an apparatus may be able to reduce the magnitude of an echo signal in a system having an audio pickup (e.g., a microphone) near an audio output (e.g., a speaker).

Abstract: An device and method of generating environmental reverberation effects for digital audio signals is presented. The device includes a reverberation controller. The reverberation controller pre-processes one or more predetermined characteristics of a first audio signal to produce a pre-processed signal and generates a plurality of delayed outputs from the pre-processed signal, each output having a predetermined delay. The reverberation controller also produces a plurality of reflection outputs from the plurality of delayed outputs and combines the plurality of reflection outputs to produce a second audio signal having a desired reverberation response.

Abstract: A system method of reusing information in a low power scalable hybrid audio encoder are disclosed. The includes determining a state of an advanced audio coding (AAC) transient flag, performing spectral band replication (SBR) transient detection on at least two possible locations upon a determination that the AAC transient flag is equal to a first value, performing SBR transient detection on a high frequency upon a determination that the AAC transient flag is equal to a second value, and determining whether a transient exists. The system includes a spectral band replication (SBR) coding module configured to determine a state of an advanced audio coding (AAC) transient flag and perform SBR transient detection on at least one location based upon an energy in a signal upon a determination that the AAC transient flag is equal to a first value.

Abstract: Methods and apparatus for self-calibration of small-microphone arrays are described. In one embodiment, self-calibration is based upon a mathematical approximation for which a detected response by one microphone should approximately equal a combined response from plural microphones in the array. In a second embodiment, self-calibration is based upon matching gains in each of a plurality of Bark frequency bands, and applying the matched gains to frequency domain microphone signals such that the magnitude response of all the microphones in the array approximates an average magnitude response for the array. The methods and apparatus may be implemented in hearing aids or small audio devices and used to mitigate adverse aging and mechanical effects on acoustic performance of small-microphone arrays in these systems.

Abstract: Audio acquisition systems and methods to determine a direction of arrival of an audio signal are disclosed. In an embodiment, an apparatus includes a continuous sampling stage configured to receive audio information and to generate one or more correlations from the received audio information, and a processing stage configured to receive the one or more correlations and to generate direction of arrival information for the audio information. In another embodiment, a method includes generating audio signals from an ambient acoustic environment, and performing beamforming on the generated audio signals. The method further includes calculating signal-to-interference ratios from the beamformed signals, forming correlations between the signal-to-interference ratios and audio sampling angles, selecting at least one correlation based upon predetermined selection criteria, and determining a direction of arrival for the audio signals.

Abstract: A time-domain method of adaptively levelling the loudness of a digital audio signal is proposed. It selects a proper frequency weighting curve to relate the volume level to the human auditory system. The audio signal is segmented into frames of a suitable duration for content analysis. Each frame is classified to one of several predefined states and events of perceptual interest is detected. Four quantities are updated each frame according to the classified state and detected event to keep track of the signal. One quantity measures the long-term loudness and is the main criterion for state classification of a frame. The second quantity is the short-term loudness that is mainly used for deriving the target gain. The third quantity measures the low-level loudness when the signal is deemed to not contain important content, giving a reasonable estimate of noise floor. A fourth quantity measures the peak loudness level that is used to simulate the temporal masking effect.