Back to results
US 8,015,002 B2Granted
Dynamic noise reduction using linear model fitting
AssigneeQNX Software Systems Limited
InventorsXueman Li, Rajeev Nongpiur, Phillip A. Hetherington
Life of this patent
Estimating the term…
Abstract
A speech enhancement system improves the speech quality and intelligibility of a speech signal. The system includes a time-to-frequency converter that converts segments of a speech signal into frequency bands. A signal detector measures the signal power of the frequency bands of each speech segment. A background noise estimator measures a background noise detected in the speech signal. A dynamic noise reduction controller dynamically models the background noise in the speech signal. The speech enhancement renders a speech signal perceptually pleasing to a listener by dynamically attenuating a portion of the noise that occurs in a portion of the spectrum of the speech signal.
Claims23
A system that improves speech quality comprising:
- a spectral converter that is configured to digitize and convert a time varying signal into the frequency domain;
- a background noise estimator configured to measure a background noise that is present in the time varying signal detected by a receiver;
- a spectral separator in communication with the spectral converter and the background noise estimator that is configured to divide a power spectrum of a speech segment;
- a modeler in communication with the spectral separator that fits a plurality of substantially linear functions to differing portions of the speech segment, where the modeler is configured to fit a first line of the plurality of substantially linear functions to a first frequency portion of the speech spectrum and a second line of the plurality of substantially linear functions to a second frequency portion of the speech spectrum;
- a dynamic noise adjuster programmed to designate the spectral magnitude of a noisy portion of the speech segment by designating a dynamic adjustment factor that corresponds to the noisy portion of the speech segment, where the dynamic noise adjuster is programmed to calculate a difference in slope between the first line and the second line, and calculate the dynamic adjustment factor based on the difference in slope;
- and
- a dynamic noise processor programmed to attenuate a portion of the noise detected in one or more portions of the speech segment based on the dynamic adjustment factor.
4 more elements
A speech enhancement system that adapts to changing noise conditions heard in a vehicle, comprising:
- a time-to-frequency converter that converts portions of a first speech segment into frequency bands;
- a signal detector configured to measure a signal power of the frequency bands of the first speech segment;
- a background noise estimator configured to measure an aural background noise detected within a vehicle;
- and
- a dynamic noise reduction controller configured to dynamically model the aural background noise in the vehicle to render an attenuated speech segment through a dynamic attenuation of a portion of the noise that occurs in a low frequency portion of the spectrum of the first speech segment below a predetermined frequency;
- where the dynamic noise reduction controller is configured to fit a first line to the low frequency portion of the spectrum of the first speech segment, fit a second line to a high frequency portion of the spectrum of the first speech segment above the predetermined frequency, calculate a difference in slope between the first line and the second line, and calculate an attenuation amount for the dynamic attenuation based on the difference in slope.
3 more elements
A system that dynamically controls the attenuation gain applied to a signal recorded in a vehicle, comprising:
- a power processor configured to measure the signal power in a sound segment in real-time;
- a background noise processor configured to measure the background noise detected in the sound segment in real-time;
- a dynamic noise reduction processor configured to model the measured background noise by processing multiple linear relationships;
- and
- a dynamic noise suppression filter having a noise suppression gain adjusted in response to the model of the measured background noise;
- where the dynamic noise reduction processor is configured to fit a first line to a first frequency portion of the sound segment, fit a second line to a second frequency portion of the sound segment, calculate a difference in slope between the first line and the second line, and calculate the noise suppression gain based on the difference in slope.
3 more elements
A method that improves speech quality and intelligibility of a speech segment, comprising:
- converting a sound segment into separate frequency bands where each band identifies an amplitude and a phase;
- estimating the background noise of a signal by averaging the acoustic power measured in each frequency band;
- discriminating between a high portion of the frequency spectrum above a predetermined frequency and a low portion of the frequency spectrum below the predetermined frequency;
- fitting a first line to the low portion of the frequency spectrum;
- fitting a second line to the high portion of the frequency spectrum;
- calculating a difference in slope between the first line and the second line;
- modeling a background noise spectrum by determining a substantially constant attenuation to be applied to the high frequency portion of the spectrum and a variable attenuation to be applied to the low portion of the frequency spectrum;
- calculating a level for the variable attenuation based on the difference in slope;
- and
- attenuating portions of the background noise from the sound segment by applying the constant attenuation and the variable attenuation.
7 more elements
A non-transitory computer readable medium that retains software that improves a speech quality by modeling a background noise comprising:
- a computer readable medium that retains a signal estimation logic, a modeling logic, and an attenuation logic that is accessible to and configured to be processed by a processor, where
- the signal estimation logic determines the signal power of a desired signal within an input signal;
- the modeling logic represents a plurality of background noises detected from the input signal through a plurality of substantially linear models, where the modeling logic fits a first line of the plurality of substantially linear models to a first frequency portion of the input signal, and fits a second line of the plurality of substantially linear models to a second frequency portion of the input signal;
- and
- the attenuation logic approximates the level of suppression to be applied to the input signal in response to an output of the modeling logic, where the attenuation logic calculates a difference in slope between the first line and the second line, and calculates the level of suppression based on the difference in slope.
2 more elements
Dependencies are inferred from the claim text.