Patents by Inventor Roman A. Dyba
Roman A. Dyba has filed for patents to protect the following inventions. This listing includes patent applications that are pending as well as patents that have already been granted by the United States Patent and Trademark Office (USPTO).
-
Patent number: 8787561Abstract: A technique for detecting in-band signaling tones in a communication system includes performing a first adaptation of an adaptive filter of an echo canceller in response to detection of a far-end harmonic signal. In this case, the adaptive filter provides an echo estimation signal. The technique also includes subtracting the echo estimation signal from a near-end signal that includes one or more in-band signaling tones to provide an error signal. The technique further includes detecting, using a tone detector, the one or more in-band signaling tones in the error signal.Type: GrantFiled: April 27, 2010Date of Patent: July 22, 2014Assignee: Freescale Semiconductor, Inc.Inventors: Roman A. Dyba, Wen Wu Su, Hongyang Deng
-
Patent number: 8693677Abstract: A technique for updating filter coefficients of an adaptive filter includes filtering a signal with an adaptive filter, whose filter coefficients are grouped into filter blocks. In this case a number of the filter blocks is less than or equal to a number of the filter coefficients. During each update period, the filter coefficients for less than all of the filter blocks are updated based on a network echo path impulse response.Type: GrantFiled: April 27, 2010Date of Patent: April 8, 2014Assignee: Freescale Semiconductor, Inc.Inventors: Hongyang Deng, Roman A. Dyba, Wen Wu Su
-
Patent number: 8666058Abstract: In one or more embodiments, an echo reduction system can include multiple adaptive filters operable to provide estimated echo replicas based on received signal information coupled to respective multiple adders that can combine a send signal and outputs of the respective adaptive filters and can provide respective output combinations to a filter selector. The filter selector can select from the outputs of the adders and provide a selected output as output for the echo reduction system. In one or more embodiments, the filter selector can control signal processing of the filters by providing control signal to the adaptive filters, where the control signals can indicate to ones of the adaptive filters to pause or continue signal processing. In one or more embodiments, the echo reduction system can include at least one delay unit that can delay receive signal information to at least one of the adaptive filters.Type: GrantFiled: September 11, 2009Date of Patent: March 4, 2014Assignee: Freescale Semiconductor, Inc.Inventors: Hongyang Deng, Roman A. Dyba
-
Patent number: 8300802Abstract: Methods and corresponding systems in an adaptive filter include calculating a signal estimator output using filter coefficients, and calculating an error signal. Next, a coefficient threshold is determined. Thereafter, for each filter coefficient, a first step size is assigned to filter coefficients with a magnitude less than the coefficient threshold, and a second step size is assigned to filter coefficients with a magnitude greater than or equal to the coefficient threshold. Finally, the filter coefficients are updated using the first and second step sizes and the error signal. The coefficient threshold can be selected as the average of the magnitudes of the filter coefficients. Alternatively, the coefficient threshold can be selected as the Mth largest of the filter coefficients ranked in order of magnitude. In one embodiment, the first step size can be less than one and the second step size can be greater than one.Type: GrantFiled: August 31, 2007Date of Patent: October 30, 2012Assignee: Freescale Semiconductor, Inc.Inventors: Hongyang Deng, Roman A. Dyba, Perry P. He
-
Patent number: 8295474Abstract: In a data processing system, a method includes receiving samples of a send path input signal (Sin) and samples of a receive path input signal (Rin); in response to receiving each sample of the send path input signal, providing a double-talk indicator to indicate whether or not double-talk has been detected; using the double-talk indicator to determine a double-talk density within a predetermined number of samples of the send path input signal or the receive path input signal; and using the double-talk density to provide a heavy double-talk indicator which indicates whether or not heavy double-talk has been detected. In response to the heavy double-talk indicator indicating that heavy double-talk has been detected, a heavy double-talk mode may be entered.Type: GrantFiled: August 8, 2008Date of Patent: October 23, 2012Assignee: Freescale Semiconductor, Inc.Inventors: Roman A. Dyba, Hongyang Deng, Brad L. Zwernemann
-
Patent number: 8290141Abstract: A technique of operating a communication device includes dividing a frequency band associated with a background noise signal into respective sub-bands. Respective individual level estimates for each of the respective sub-bands are then determined. A total level estimate for the background noise signal is determined. Finally, a comfort noise signal (whose characteristics are based on the respective individual level estimates and the total level estimate) is provided.Type: GrantFiled: April 18, 2008Date of Patent: October 16, 2012Assignee: Freescale Semiconductor, Inc.Inventors: Roman A. Dyba, Perry P. He, Brad L. Zwernemann
-
Patent number: 8139760Abstract: Methods and corresponding systems for reducing an echo signal include assigning a subsegment delay to each of a plurality of subsegment adaptive filters. A send signal and a delayed receive signal are coupled to each of the subsegment adaptive filters, wherein the delayed receive signal is delayed according to a respective subsegment delay. A set of filter coefficients in each of the subsegment adaptive filters are adapted, in parallel, to correspond to a respective subsegment impulse response of a connected system. Each set of filter coefficients is analyzed to produce a pure delay, and the pure delay is used to delay the receive signal for a main adaptive filter. An echo signal replica is produced using the main adaptive filter and an error signal. The echo signal replica is subtracted from the send signal to produce the error signal.Type: GrantFiled: February 27, 2007Date of Patent: March 20, 2012Assignee: Freescale Semiconductor, Inc.Inventors: Roman A. Dyba, Perry P. He
-
Publication number: 20110261948Abstract: A technique for updating filter coefficients of an adaptive filter includes filtering a signal with an adaptive filter, whose filter coefficients are grouped into filter blocks. In this case a number of the filter blocks is less than or equal to a number of the filter coefficients. During each update period, the filter coefficients for less than all of the filter blocks are updated based on a network echo path impulse response.Type: ApplicationFiled: April 27, 2010Publication date: October 27, 2011Applicant: FREESCALE SEMICONDUCTOR, INC.Inventors: HONGYANG DENG, Roman A. Dyba, Wen Wu Su
-
Publication number: 20110261949Abstract: A technique for detecting in-band signaling tones in a communication system includes performing a first adaptation of an adaptive filter of an echo canceller in response to detection of a far-end harmonic signal. In this case, the adaptive filter provides an echo estimation signal. The technique also includes subtracting the echo estimation signal from a near-end signal that includes one or more in-band signaling tones to provide an error signal. The technique further includes detecting, using a tone detector, the one or more in-band signaling tones in the error signal.Type: ApplicationFiled: April 27, 2010Publication date: October 27, 2011Applicant: FREESCALE SEMICONDUCTOR, INC.Inventors: Roman A. Dyba, Wen Wu Su, Hongyang Deng
-
Publication number: 20110064213Abstract: In one or more embodiments, an echo reduction system can include multiple adaptive filters operable to provide estimated echo replicas based on received signal information coupled to respective multiple adders that can combine a send signal and outputs of the respective adaptive filters and can provide respective output combinations to a filter selector. The filter selector can select from the outputs of the adders and provide a selected output as output for the echo reduction system. In one or more embodiments, the filter selector can control signal processing of the filters by providing control signal to the adaptive filters, where the control signals can indicate to ones of the adaptive filters to pause or continue signal processing. In one or more embodiments, the echo reduction system can include at least one delay unit that can delay receive signal information to at least one of the adaptive filters.Type: ApplicationFiled: September 11, 2009Publication date: March 17, 2011Applicant: FREESCALE SEMICONDUCTOR, INC.Inventors: Hongyang Deng, Roman A. Dyba
-
Publication number: 20110013766Abstract: A system includes receive and send inputs, an adaptive filter, a combiner, a non-linear processor, and a tone detector. The receive input receives a first signal having first voice and first tone information. The send input receives a second signal having echo information, second tone information, and second voice information. The adaptive filter has an input coupled to the receive input, and an output. The combiner has a first input coupled to the send input, a second input coupled to the output of the adaptive filter, and an output providing an intermediate signal. The nonlinear processor has an input coupled to the output of the combiner and an output for providing an output signal for obtaining the second voice. The tone detector is coupled to the output of the combiner and detects the second tone in the intermediate signal prior to the intermediate signal being processed by the nonlinear processor.Type: ApplicationFiled: July 15, 2009Publication date: January 20, 2011Inventors: Roman A. Dyba, Hongyang Deng, Wen Wu Su
-
Publication number: 20100034373Abstract: In a data processing system, a method includes receiving samples of a send path input signal (Sin) and samples of a receive path input signal (Rin); in response to receiving each sample of the send path input signal, providing a double-talk indicator to indicate whether or not double-talk has been detected; using the double-talk indicator to determine a double-talk density within a predetermined number of samples of the send path input signal or the receive path input signal; and using the double-talk density to provide a heavy double-talk indicator which indicates whether or not heavy double-talk has been detected. In response to the heavy double-talk indicator indicating that heavy double-talk has been detected, a heavy double-talk mode may be entered.Type: ApplicationFiled: August 8, 2008Publication date: February 11, 2010Inventors: ROMAN A. DYBA, Hongyang Deng, Brad L. Zwernemann
-
Patent number: 7639671Abstract: Methods and corresponding systems for allocating processing resources for a number of instances (N) of a software component include determining an average processing cost (?) and a variance (?2) for the software component. Then a processing cost for the software component is estimated as a function of N, the average processing cost (?), and the variance (?2), and processing resources are allocated in response to the estimated processing cost. The software component can be partitioned into a number of blocks (L), wherein the L blocks include a required block and one or more optional blocks. In some embodiments in response to a total estimated processing cost exceeding an available processing value, selected optional blocks can be disabled to reduce the total estimated processing cost to a value equal to or less than the available processing value. The optional blocks can be prioritized and disabled in order of priority.Type: GrantFiled: March 7, 2006Date of Patent: December 29, 2009Assignee: Freescale Semiconductor, Inc.Inventors: Brad L. Zwernemann, Roman A. Dyba, Perry P. He, Lucio F. C. Pessoa
-
Publication number: 20090265169Abstract: A technique of operating a communication device includes dividing a frequency band associated with a background noise signal into respective sub-bands. Respective individual level estimates for each of the respective sub-bands are then determined. A total level estimate for the background noise signal is determined. Finally, a comfort noise signal (whose characteristics are based on the respective individual level estimates and the total level estimate) is provided.Type: ApplicationFiled: April 18, 2008Publication date: October 22, 2009Inventors: Roman A. Dyba, Perry P. He, Brad L. Zwernemann
-
Patent number: 7599432Abstract: A method and apparatus for dynamically inserting gain in an adaptive filter is taught. This selective insertion of gain may be used to allow an adaptive filter to converge more quickly and/or to overcome inherent limitations of the adaptive filter. An echo canceller (e.g. 20 and 22 in FIG. 1) is just one possible use for an adaptive filter having a shorter convergence time. However, virtually all uses of an adaptive filter can benefit from a shorter convergence time and/or improved filtering performance.Type: GrantFiled: December 8, 2003Date of Patent: October 6, 2009Assignee: Freescale Semiconductor, Inc.Inventors: Lucio F. C. Pessoa, Roman A. Dyba, Perry P. He, Ahsan Ul Aziz
-
Patent number: 7555075Abstract: Methods and corresponding systems for suppressing noise in an input signal include setting a minimum overall gain in a noise reduction processor for processing a first frame of data associated with the input signal. In response to a new minimum overall gain being set, the minimum overall gain in the noise reduction processor is replaced with the new minimum overall gain, and a second frame of data associated with the input signal is processed to suppress noise using the new minimum overall gain. The new minimum overall gain can be a function of the input signal or an output signal of the noise reduction processor. The new minimum overall gain can correspond to a difference between an estimated signal-to-noise ratio (SNR) improvement that is calculated using time-domain data and a target SNR improvement.Type: GrantFiled: April 7, 2006Date of Patent: June 30, 2009Assignee: Freescale Semiconductor, Inc.Inventors: Lucio F. C. Pessoa, Roman A. Dyba, David B. Melles
-
Publication number: 20090060167Abstract: Methods and corresponding systems in an adaptive filter include calculating a signal estimator output using filter coefficients, and calculating an error signal. Next, a coefficient threshold is determined. Thereafter, for each filter coefficient, a first step size is assigned to filter coefficients with a magnitude less than the coefficient threshold, and a second step size is assigned to filter coefficients with a magnitude greater than or equal to the coefficient threshold. Finally, the filter coefficients are updated using the first and second step sizes and the error signal. The coefficient threshold can be selected as the average of the magnitudes of the filter coefficients. Alternatively, the coefficient threshold can be selected as the Mth largest of the filter coefficients ranked in order of magnitude. In one embodiment, the first step size can be less than one and the second step size can be greater than one.Type: ApplicationFiled: August 31, 2007Publication date: March 5, 2009Inventors: Hongyang Deng, Roman A. Dyba, Perry P. He
-
Publication number: 20080205633Abstract: Methods and corresponding systems for reducing an echo signal include assigning a subsegment delay to each of a plurality of subsegment adaptive filters. A send signal and a delayed receive signal are coupled to each of the subsegment adaptive filters, wherein the delayed receive signal is delayed according to a respective subsegment delay. A set of filter coefficients in each of the subsegment adaptive filters are adapted, in parallel, to correspond to a respective subsegment impulse response of a connected system. Each set of filter coefficients is analyzed to produce a pure delay, and the pure delay is used to delay the receive signal for a main adaptive filter. An echo signal replica is produced using the main adaptive filter and an error signal. The echo signal replica is subtracted from the send signal to produce the error signal.Type: ApplicationFiled: February 27, 2007Publication date: August 28, 2008Inventors: Roman A. Dyba, Perry P. He
-
Patent number: 7388954Abstract: A communication system having an echo canceller is disclosed. One embodiment of the echo canceller includes an adaptive filter used to provide an estimate of reflected echo which is removed from the send signal. The echo canceller may also include a near-end talker signal detector which may be used to prevent the adaptive filter from adapting when a near-end talker signal is present. The echo canceller may also include a nonlinear processor used to further reduce any residual echo and to preserve background noise. The echo canceller may also include a monitor and control unit which may be used to monitor the filter coefficients and gain of the adaptive filter to maintain stability of the echo canceller, estimate pure delay, detect a tone, and inject a training signal. The echo canceller may also include a nonadaptive filter used to reduce the length of the adaptive filter.Type: GrantFiled: June 24, 2002Date of Patent: June 17, 2008Assignee: Freescale Semiconductor, Inc.Inventors: Lucio F. C. Pessoa, Roman A. Dyba, Perry P. He
-
Publication number: 20070237271Abstract: Methods and corresponding systems for suppressing noise in an input signal include setting a minimum overall gain in a noise reduction processor for processing a first frame of data associated with the input signal. In response to a new minimum overall gain being set, the minimum overall gain in the noise reduction processor is replaced with the new minimum overall gain, and a second frame of data associated with the input signal is processed to suppress noise using the new minimum overall gain. The new minimum overall gain can be a function of the input signal or an output signal of the noise reduction processor. The new minimum overall gain can correspond to a difference between an estimated signal-to-noise ratio (SNR) improvement that is calculated using time-domain data and a target SNR improvement.Type: ApplicationFiled: April 7, 2006Publication date: October 11, 2007Inventors: Lucio Pessoa, Roman Dyba, David Melles