Patent US2602758A successfully refreshed from Google Patents.
US6163608A
Expired / Lapsed
US
⏱️ 0.0 years remaining
Methods and apparatus for providing comfort noise in communications systems
Assignee: Ericsson Inc
•
Filing: 1998-01-09
•
Granted: 2000-12-19
•
Expires: 2018-01-09
Abstract
Abstract Methods and apparatus for parametrically modeling background noise and generating comfort noise in echo suppression systems are disclosed. According to exemplary embodiments, a background noise model is based on a set of noise model parameters which are in turn based on measurements of actual background noise in an echo suppression system. The exemplary embodiments include autoregressive, autoregressive moving-average and frequency-domain models. An exemplary first-order autoregressive moving-average model includes a single fixed zero and a single variable pole. The single zero and the single pole are sufficient to provide appropriate spectral tilt in the resulting modeled noise, and the single zero ensures that the model is unconditionally stable. Integration of the exemplary parametric noise models with various echo suppression devices is also disclosed.
Inventors
- 👤 Eric Douglas Romesburg
- 👤 Leland Scott Bloebaum
- 👤 Corattur Natesan Sambandam Guruparan
Bibliographic Details
Application Number
—
Jurisdiction
US
Data Source
google_patents
Last Synchronized
2026-11-05 08:57
CPC Classifications
H; H04; H04M; H04M9/00; H04M9/08; H04M9/082
Claims Analysis
Total Claims: 55
Claim 1
Independent Claim
74 words
An echo suppression device configured to attenuate an echo component of a communications signal, said device comprising: a noise modeling processor configured to generate at least one noise modeling parameter based upon said communications signal, said at least one noise modeling parameter defining a parametric model of a noise component of said communications signal; and a noise generation processor configured to provide modeled noise samples based upon said at least one noise modeling parameter.
Claim 2
Dependent Claim
74 words
An echo suppression device configured to attenuate an echo component of a communications signal, said device comprising: a noise modeling processor configured to generate at least one noise modeling parameter based upon said communications signal, said at least one noise modeling parameter defining a parametric model of a noise component of said communications signal; and a noise generation processor configured to provide modeled noise samples based upon said at least one noise modeling parameter.
Claim 3
Dependent Claim
46 words
The echo suppression device according to claim 1, wherein said echo suppression device attenuates both echo and noise components of said communications signal to provide an output signal, and wherein said modeled noise samples are added to said output signal to replace the attenuated noise component.
Claim 4
Dependent Claim
23 words
The echo suppression device according to claim 1, wherein the noise component of said communications signal is modeled as an autoregressive random process.
Claim 5
Dependent Claim
39 words
The echo suppression device according to claim 3, wherein said noise modeling processor is configured to generate said at least one noise modeling parameter by computing an autocorrelation vector based upon a frame of samples of said communications signal.
Claim 6
Dependent Claim
28 words
The echo suppression device according to claim 4, wherein said noise modeling processor is configured to recursively smooth autocorrelation vectors computed for multiple frames of said communications signal.
Claim 7
Dependent Claim
28 words
The echo suppression device according to claim 4, wherein said noise modeling processor is configured to non-recursively smooth autocorrelation vectors computed for multiple frames of said communications signal.
Claim 8
Dependent Claim
39 words
The echo suppression device according to claim 4, wherein said noise generation processor is configured to compute a set of filter coefficients based upon said autocorrelation vector, said filter coefficients defining a spectral shaping filter for said autoregressive process.
Claim 9
Dependent Claim
18 words
The echo suppression device according to claim 7, wherein said filter coefficients are computed using a Levinson-Durbin algorithm.
Claim 10
Dependent Claim
51 words
The echo suppression device according to claim 8, wherein said noise generation processor is configured to generate a zero-mean pseudo-random sequence, said sequence having a variance proportional to a zero-lag coefficient of the autocorrelation vector, and to filter said sequence using said spectral shaping filter to provide said modeled noise samples.
Claim 11
Dependent Claim
56 words
The echo suppression device according to claim 3, wherein said noise modeling processor is configured to generate said at least one noise modeling parameter by using a recursive predictive algorithm to compute a set of filter coefficients based on samples of said communications signal, said filter coefficients defining a spectral shaping filter for said autoregressive process.
Claim 12
Dependent Claim
18 words
The echo suppression device according to claim 10, wherein said adaptive algorithm is a Least Mean Squares algorithm.
Claim 13
Dependent Claim
19 words
The echo suppression device according to claim 10, wherein said adaptive algorithm is a Normalized Least Mean Squares algorithm.
Claim 14
Dependent Claim
56 words
The echo suppression device according to claim 10, wherein said noise generation processor is configured to generate a zero-mean pseudo-random sequence, said sequence having a variance proportional to an energy level of a residual error signal of said adaptive algorithm, and to filter said sequence using said spectral shaping filter to provide said modeled noise samples.
Claim 15
Dependent Claim
23 words
The echo suppression device according to claim 1, wherein the noise component of said communications signal is modeled as an autoregressive moving-average process.
Claim 16
Dependent Claim
44 words
The echo suppression device according to claim 14, wherein a set of zeros in a spectral shaping filter defining said autoregressive moving-average process are fixed based upon a priori information relating to an environment in which said echo suppression device is to be implemented.
Claim 17
Dependent Claim
39 words
The echo suppression device according to claim 15, wherein said noise modeling processor is configured to generate said at least one noise modeling parameter by computing an autocorrelation vector based upon a frame of samples of said communications signal.
Claim 18
Dependent Claim
46 words
The echo suppression device according to claim 16, wherein said frame of samples is filtered by an intermediate filter before said autocorrelation vector is computed, said intermediate filter including a set of fixed poles corresponding to the set of fixed zeros of said spectral shaping filter.
Claim 19
Dependent Claim
28 words
The echo suppression device according to claim 16, wherein said noise modeling processor is configured to recursively average autocorrelation vectors computed for multiple frames of said communications signal.
Claim 20
Dependent Claim
28 words
The echo suppression device according to claim 16, wherein said noise modeling processor is configured to non-recursively smooth autocorrelation vectors computed for multiple frames of said communications signal.
Claim 21
Dependent Claim
31 words
The echo suppression device according to claim 16, wherein said noise generation processor is configured to compute a set of poles for said spectral shaping filter based upon said autocorrelation vector.
Claim 22
Dependent Claim
17 words
The echo suppression device according to claim 20, wherein said poles are computed using a Levinson-Durbin algorithm.
Claim 23
Dependent Claim
51 words
The echo suppression device according to claim 20, wherein said noise generation processor is configured to generate a zero-mean pseudo-random sequence, said sequence having a variance proportional to a zero-lag coefficient of the autocorrelation vector, and to filter said sequence using said spectral shaping filter to generate said modeled noise samples.
Claim 24
Dependent Claim
48 words
The echo suppression device according to claim 15, wherein said noise modeling processor is configured to generate said at least one noise modeling parameter by using a recursive predictive algorithm to compute a set of poles for said spectral shaping filter based upon samples of said communications signal.
Claim 25
Dependent Claim
18 words
The echo suppression device according to claim 23, wherein said adaptive algorithm is a Least Mean Squares algorithm.
Claim 26
Dependent Claim
19 words
The echo suppression device according to claim 23, wherein said adaptive algorithm is a Normalized Least Mean Squares algorithm.
Claim 27
Dependent Claim
56 words
The echo suppression device according to claim 23, wherein said noise generation processor is configured to generate a zero-mean pseudo-random sequence, said sequence having a variance proportional to an energy level of a residual error signal of said adaptive algorithm, and to filter said sequence using said spectral shaping filter to provide said modeled noise samples.
Claim 28
Dependent Claim
38 words
The echo suppression device according to claim 14, wherein said autoregressive moving average process is a first-order process, and wherein a spectral shaping filter defining said first-order process includes a single fixed zero and a single variable pole.
Claim 29
Dependent Claim
34 words
The echo suppression device according to claim 27, wherein said single fixed zero is set based upon a priori information relating to an environment in which said echo suppression device is to be implemented.
Claim 30
Dependent Claim
18 words
The echo suppression device according to claim 27, wherein said single fixed zero is set to approximately -1.
Claim 31
Dependent Claim
17 words
The echo suppression device according to claim 27, wherein said single fixed zero is set to -13/16.
Claim 32
Dependent Claim
30 words
The echo suppression device according to claim 27, wherein said single variable pole is given by a, and wherein a gain of said spectral shaping filter is proportional to 1-a.
Claim 33
Dependent Claim
33 words
The echo suppression device according to claim 27, wherein said noise modeling processor is configured to use an adaptive algorithm to compute said single variable pole based upon samples of said communications signal.
Claim 34
Dependent Claim
18 words
The echo suppression device according to claim 32, wherein said adaptive algorithm is a Least Mean Squares algorithm.
Claim 35
Dependent Claim
19 words
The echo suppression device according to claim 32, wherein said adaptive algorithm is a Normalized Least Mean Squares algorithm.
Claim 36
Dependent Claim
56 words
The echo suppression device according to claim 32, wherein said noise generation processor is configured to generate a zero-mean pseudo-random sequence, said sequence having a variance proportional to an energy level of a residual error signal of said adaptive algorithm, and to filter said sequence using said spectral shaping filter to provide said modeled noise samples.
Claim 37
Dependent Claim
32 words
The echo suppression device according to claim 1, wherein said model of the noise component of said communications signal is based upon a linear orthogonal transformation of samples of said communications signal.
Claim 38
Dependent Claim
23 words
The echo suppression device according to claim 1, wherein said model of the noise component of said communications signal is a frequency-domain model.
Claim 39
Dependent Claim
41 words
The echo suppression device according to claim 37, wherein said noise modeling processor is configured to generate said at least one noise modeling parameter by computing a vector of spectral magnitudes based upon a frame of samples of said communications signal.
Claim 40
Dependent Claim
29 words
The echo suppression device according to claim 38, wherein said spectral magnitude vector is derived by computing a Fourier transform of said frame of samples of said communications signal.
Claim 41
Dependent Claim
29 words
The echo suppression device according to claim 38, wherein said noise modeling processor is configured to recursively smooth spectral magnitude vectors computed for multiple frames of said communications signal.
Claim 42
Dependent Claim
29 words
The echo suppression device according to claim 38, wherein said noise modeling processor is configured to non-recursively smooth spectral magnitude vectors computed for multiple frames of said communications signal.
Claim 43
Dependent Claim
39 words
The echo suppression device according to claim 38, wherein said noise generation processor is configured to generate a pseudo-random sequence of phase values and to compute said modeled noise samples based upon said sequence and said spectral magnitude vector.
Claim 44
Dependent Claim
52 words
The echo suppression device according to claim 42, wherein said modeled noise samples are derived by computing an inverse Fourier transform of a complex vector, each complex sample in said complex vector including a spectral magnitude from said spectral magnitude vector and a pseudo-random phase from said sequence of pseudo-random phase values.
Claim 45
Dependent Claim
60 words
The echo suppression device according to claim 1, wherein said noise generation processor is configured to pseudo-randomly select single samples from a buffer of samples of said communications signal and to filter a sequence of said pseudo-randomly selected single samples, using a spectral shaping filter defined by said at least one noise modeling parameter, to provide said modeled noise samples.
Claim 46
Dependent Claim
69 words
An echo suppression device, comprising: an echo suppressor configured to attenuate echo and noise components of a communications signal; a sample buffer for storing frames of samples of said communications signal; and a noise generation processor configured to pseudo-randomly select single samples from said buffer in order to provide a sequence of white noise samples having a power level equal to a power level of said frame of samples.
Claim 47
Dependent Claim
69 words
An echo suppression device, comprising: an echo suppressor configured to attenuate echo and noise components of a communications signal; a sample buffer for storing frames of samples of said communications signal; and a noise generation processor configured to pseudo-randomly select single samples from said buffer in order to provide a sequence of white noise samples having a power level equal to a power level of said frame of samples.
Claim 48
Dependent Claim
34 words
The echo suppression device according to claim 45, wherein said noise generation processor filters said sequence of white noise samples to provide comfort noise which is added to an output of said echo suppressor.
Claim 49
Dependent Claim
149 words
An echo suppression device, comprising: an echo suppressor configured to attenuate echo and noise components of a communications signal; a noise modeling and generation processor configured to model the noise component of said communications signal and to provide comfort noise for said echo suppression device based on said model; a voice activity detector configured to provide an indication of whether said communications signal includes a voice component; and a noise level estimator configured to compute an estimate of a noise level of said communications signal and to provide an indication of whether an energy level of said communications signal is less than said estimate, wherein said noise modeling and generation processor is configured to update said model only when said voice activity detector indicates that there is no voice component in the communications signal and said noise level estimator indicates that said energy level is less than said estimate.
Claim 50
Dependent Claim
149 words
An echo suppression device, comprising: an echo suppressor configured to attenuate echo and noise components of a communications signal; a noise modeling and generation processor configured to model the noise component of said communications signal and to provide comfort noise for said echo suppression device based on said model; a voice activity detector configured to provide an indication of whether said communications signal includes a voice component; and a noise level estimator configured to compute an estimate of a noise level of said communications signal and to provide an indication of whether an energy level of said communications signal is less than said estimate, wherein said noise modeling and generation processor is configured to update said model only when said voice activity detector indicates that there is no voice component in the communications signal and said noise level estimator indicates that said energy level is less than said estimate.
Claim 51
Dependent Claim
84 words
An echo suppression device, comprising: an echo suppressor configured to attenuate echo and noise components of a communications signal, wherein said echo suppressor removes a portion of said communications signal falling within an amplitude attenuation window; and a comfort noise generator configured to provide comfort noise for said echo suppression device, wherein an output of said comfort noise generator is limited to said amplitude attenuation window to provide a limited comfort noise output which is added to an audio output of said echo suppressor.
Claim 52
Dependent Claim
84 words
An echo suppression device, comprising: an echo suppressor configured to attenuate echo and noise components of a communications signal, wherein said echo suppressor removes a portion of said communications signal falling within an amplitude attenuation window; and a comfort noise generator configured to provide comfort noise for said echo suppression device, wherein an output of said comfort noise generator is limited to said amplitude attenuation window to provide a limited comfort noise output which is added to an audio output of said echo suppressor.
Claim 53
Dependent Claim
111 words
An echo suppression device, comprising: an echo suppressor configured to attenuate echo and noise components of a communications signal, wherein said echo suppressor multiplies a portion of said communications signal falling within an amplitude attenuation window by a scale factor; and a comfort noise generator configured to provide comfort noise for said echo suppression device, wherein an output of said comfort noise generator is limited to said amplitude attenuation window to provide a limited comfort noise output, and wherein said limited comfort noise output is scaled based on said scale factor to provide a limited and scaled comfort noise output which is added to an audio output of said echo suppressor.
Claim 54
Dependent Claim
111 words
An echo suppression device, comprising: an echo suppressor configured to attenuate echo and noise components of a communications signal, wherein said echo suppressor multiplies a portion of said communications signal falling within an amplitude attenuation window by a scale factor; and a comfort noise generator configured to provide comfort noise for said echo suppression device, wherein an output of said comfort noise generator is limited to said amplitude attenuation window to provide a limited comfort noise output, and wherein said limited comfort noise output is scaled based on said scale factor to provide a limited and scaled comfort noise output which is added to an audio output of said echo suppressor.
Claim 55
Dependent Claim
29 words
The echo suppression device according to claim 45, wherein the single samples are selected in such a manner that prevents sequential replication of substantial portions of the sample buffer.
References Cited (Backward Citations)
US5283784A
Cited
GB2256351A
Cited
US5630016A
Cited
US5475731A
Cited
EP0704988A2
Cited
WO1996022651A1
Cited
WO1996042142A1
Cited
WO1997049196A2
Cited
USPTO Annuity & Maintenance Windows
| Annuity Stage | Window Opens | Payment Deadline | Surcharge Period Until | Estimated Fee | Status |
|---|---|---|---|---|---|
| 3.5 Years (Year 4) | 2003-12-19 | 2004-06-19 | 2004-12-19 | $2000.00 | Window Closed / Lapsed |
| 7.5 Years (Year 8) | 2007-12-19 | 2008-06-19 | 2008-12-19 | $3760.00 | Window Closed / Lapsed |
| 11.5 Years (Year 12) | 2011-12-19 | 2012-06-19 | 2012-12-19 | $7700.00 | Window Closed / Lapsed |