Audience, Inc.

Interval Research Corporation

Applied Neurosystems Corporation

Lloyd Watts

Peter Santos

Carlos M Avendano

David Klein

Carlo Murgia

Ludger Solbach

Brian D Clark

Stephen Malinowski

Bernard Mont-Reynaud

Carver Mead

Forest Baskett

Provided are systems and methods for image enhancement based on combining multiple related images, such as images of the same object taken from different imaging angles. This approach allows simulating images captured from longer distances using telephoto lenses. Initial images may be captured using simple cameras equipped with shorter focal length lenses, typically used on camera phones, tablets, and laptops. The initial images may be taken using two different cameras positioned a certain distance from each other. An object or, more specifically, a center line of the object is identified in each image. The object is typically present in the foreground portion of the initial images. The initial images may be cross-faded along the object center line to yield a combined image. Separating of the foreground and background portions of each image may be separated and separately processed, such as blurring the background portion and sharpening the foreground portion.

Image enhancement based on combining images from multiple cameras

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.

System and method for utilizing inter-microphone level differences for speech enhancement

Provided are computer-implemented systems and methods for image correction after combining images from multiple cameras. While combining multiple images may significantly enhance image quality without a need for a longer focal length lens, combined images may have some perspective distortion caused by different imaging angles used by different cameras. For example, when two cameras provided on the same device are used to capture images of a face and these images are combined by cross-fading, a combined image may have an unnaturally widened center portion. Specifically, proportions of nose and ears may be distorted resulting in a wider nose and generally a wider face. The combined image may be compressed either uniformly or non-uniformly (e.g., more compression in the center) to yield a processed image with more realistic and/or natural proportions. The compression ratio and/or compression distribution may depend on imaging angles, various distances, and other factors described herein.

Image correction after combining images from multiple cameras

Background audio can be provided during telephonic communication. Telephonic communication can be established via a network, such as between a user of a telephony device and a communication partner having a second telephony device. A voice signal may be received from the user via a microphone integral with the telephony device. An audio track can be retrieved, for example, from memory integral with the telephony device or from a third-party service provider via the communications network. Noise reduction is performed on the voice signal to produce a clean voice signal. The clean voice signal may be combined with the audio track to produce a combined signal, such that the audio track provides background audio to the clean voice signal. The combined signal can then be transmitted from the telephony device to the second telephony device via the communications network.

Providing background audio during telephonic communication

Systems and methods for modification of an audio input signal are provided. In exemplary embodiments, an adaptive multiple-model optimizer is configured to generate at least one source model parameter for facilitating modification of an analyzed signal. The adaptive multiple-model optimizer comprises a segment grouping engine and a source grouping engine. The segment grouping engine is configured to group simultaneous feature segments to generate at least one segment model. The at least one segment model is used by the source grouping engine to generate at least one source model, which comprises the at least one source model parameter. Control signals for modification of the analyzed signal may then be generated based on the at least one source model parameter.

Systems and methods for audio signal analysis and modification

Systems and methods for audio signal processing are provided. In exemplary embodiments, a filter cascade of complex-valued filters are used to decompose an input audio signal into a plurality of frequency components or sub-band signals. These sub-band signals may be processed for phase alignment, amplitude compensation, and time delay prior to summation of real portions of the sub-band signals to generate a reconstructed audio signal.

System and method for processing an audio signal

The present technology measures distortion introduced by a noise suppression system. The distortion may be measured as the difference between a noise-reduced speech signal and an estimated idealized noise reduced reference (EINRR). The EINRR may be determined from a speech component and noise component that are pre-processed, and the EINRR may be used with masks associated with energies lost and added in the speech component and noise component. The EINRR may be calculated on a time varying basis.

Distortion measurement for noise suppression system

A method of removing reverberation from audio signals is disclosed. The method comprises spectro-temporally analyzing the first audio signal and the second audio signal to derive an energy function of time for a plurality of frequency bands. The method further comprises determining a delay stability between the energy function of time for the first audio signal and the second audio signal in each band, determining a gain function in each band based on the delay stability, adjusting the energy of the first audio signal and the second audio signal using the gain function within each band, and resynthesizing audio signals from the energy in each band of the first audio signal and the second audio signal.

Reverberation removal

A system and method are disclosed for processing an audio signal including separating the audio signal into a plurality of streams which group sounds from a same source prior to classification and analyzing each separate stream to determine phoneme-level classification. One or more words of the audio signal may then be outputted.

Multi-source phoneme classification for noise-robust automatic speech recognition

A system for processing audio signals comprises a sequence of digital filters each configured to process a selected frequency using a set of coefficients. A filter configured to process a certain frequency shares its coefficients with another filter that processes a frequency that is lower than the first frequency by at least one frequency interval, such as an octave. The first filter samples at a certain sampling rate, and the second filter's sampling rate is determined by multiplying the first sampling rate by the ratio of the second frequency to the first frequency. The filters are evenly grouped into frequency intervals, such as octaves. Filters in an octave are sampled at a sampling frequency that is at least twice as high as the highest frequency processed in that octave.

Efficient computation of log-frequency-scale digital filter cascade

Determining a time delay between a first signal received at a first sensor and a second signal received at a second sensor is described. The first signal is analyzed to derive a plurality of first signal channels at different frequencies and the second signal is analyzed to derive a plurality of second signal channels at different frequencies. A first feature detected that occurs at a first time in one of the first signal channels. A second feature is detected that occurs at a second time in one of the second signal channels. The first feature is matched with the second feature and the first time is compared to the second time to determine the time delay.

Computation of multi-sensor time delays

A memory cell having a memory cell state is disclosed. The memory cell state includes a memory unit configured to store a bit of data wherein the bit of data determines whether the memory cell state is a first state or a second state and a common state indicator line configured to indicate a common state associated with a group of memory cells. The memory cell state further comprises a logic circuit configured to compare the memory cell state to the common state and an output line is configured to indicate the result of the comparison. A method of identifying one or more minimum value elements in a set of elements is also disclosed. The method includes storing the values in a memory cell array having a plurality of sets of memory cells, each value being stored in a set of memory cells. The stored values are compared to a set of common state indicator lines configured to indicate the minimum value stored in the memory cell array. The method further includes noting which sets of memory cells store the minimum value.

Growing community of inventors

Mountain View, CA, United States of America

Lloyd Watts