# US 2019/0318733 A1 · Adaptive enhancement of speech signals > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 2019/0318733 A1 - Document ID: US20190318733A1 - Status: Published application - Kind: A1 - Application number: 15/951,708 - Filed: 2018-04-12 - Published: 2019-10-17 - Earliest claimed: 2018-04-12 - Assignees: Kaam LLC - Applicants: Kaam LLC. - Inventors: SENTHIL KUMAR MANI - Web page: https://patentgrep.com/patent/US20190318733A1 ## CPC - G10L 15/20: Speech recognition techniques specially adapted for robustness in adverse environments, e.g. in noise, of stress induced speech (G10L21/02 takes precedence) - G10L 21/0208: Noise filtering - G10L 15/22: Procedures used during a speech recognition process, e.g. man-machine dialogue - G10L 21/0216: characterised by the method used for estimating noise - G10L 25/84: for discriminating voice from noise - G10L 25/15: the extracted parameters being formant information - G10L: SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING ## Abstract A signal processing apparatus that handles an adaptive enhancement of a speech signal, receives a first signal and a second signal from a determined source. At least one of a speech signal or at least one noise signal is present in the first signal or the second signal. The first signal and the received second signal are processed to obtain a processed signal for amplification of a gain associated with the speech signal present in the first signal and the second signal by a determined factor. A signal-to-noise ratio (SNR) associated with the processed signal is greater than or equal to a threshold value. A reference noise signal is obtained from the second signal based on subtraction of an estimated the speech signal present in the received second signal from the processed signal. A processed speech signal is determined based on filtration of the obtained reference noise signal. ## Claims ### Claim 1 A system, comprising: a plurality of circuits in a signal processing apparatus, wherein the plurality of circuits are configured to: receive a first signal and a second signal from determined sources, wherein the first signal comprises at least one of a speech signal or at least one noise signal, and wherein the second signal comprises at least one of the speech signal or the at least one noise signal; obtain a processed signal based on the received first signal and the received second signal for amplification of a gain associated with the speech signal present in the received first signal and the second signal by a determined factor, wherein a signal-to-noise ratio (SNR) associated with the processed signal is greater than or equal to a threshold value; obtain a reference noise signal from the received second signal based on subtraction of an estimated the speech signal present in the received second signal from the processed signal; and determine a processed speech signal based on filtration of the obtained reference noise signal. ### Claim 2 (depends on 1) The system according to claim 1, wherein at least one circuit of the plurality of circuits is configured to: estimate a component of noise corresponding to the at least one noise signal in the received second signal; filter the estimated component of noise from the received second signal; and obtain a filtered second signal that comprises at least one of the speech signal or a reverberated speech signal. ### Claim 3 (depends on 2) The system according to claim 2, wherein at least one circuit of the plurality of circuits is configured to: obtain a single speech signal based on modification of the at least one of the speech signal or the reverberated speech signal in the filtered second signal, wherein the single speech signal is associated with a phase and a gain that matches with corresponding phase and gain of the speech signal in the received first signal; and obtain the processed signal based on constructive combination of the obtained single speech signal and the received first signal, wherein the processed signal is associated with a gain amplified by the determined factor. ### Claim 4 (depends on 1) The system according to claim 1, wherein at least one circuit of the plurality of circuits is configured to determine a noise path impulse response corresponding to a noise path and a speech path impulse response corresponding to a speech path, wherein the noise path corresponds to a path of reception of the at least one noise signal, and wherein the speech path corresponds to a path of reception of the speech signal. ### Claim 5 (depends on 1) The system according to claim 1, wherein at least one circuit of the plurality of circuits is configured to: determine each formant in a set of formants in the speech signal in the first signal, wherein each formant corresponds to a resonant frequency bin in the speech signal; identify a defined number of frequency bins of a defined window size adjacent to each determined formant in the set of formants in a specified pattern for adaptation to the speech signal in the received first signal; and estimate an adapted step-size corresponding to each determined formant and each frequency bin of the identified defined number of frequency bins in the speech signal present in the received first signal; and filter the speech signal from the first signal based on the estimated step-size. ### Claim 6 (depends on 1) The system according to claim 1, wherein at least one circuit of the plurality of circuits is configured to implement a sparseness controlled improved proportionate multiline delay filtering technique for determination of each formant and corresponding frequency bins in the speech signal of the received first signal. ### Claim 7 (depends on 1) The system according to claim 1, wherein the determined factor by which the gain is amplified is 2. ### Claim 8 A method, comprising: in a signal processing apparatus that comprises a plurality of circuits: receiving, by at least one circuit of the plurality of circuits, a first signal and a second signal from a determined source, wherein the first signal comprises at least one of a speech signal or at least one noise signal, and wherein the second signal comprises at least one of the speech signal or the at least one noise signal; obtaining, by at least one circuit of the plurality of circuits, a processed signal based on the received first signal and the received second signal for amplification of a gain associated with the speech signal present in the received first signal and the second signal by a determined factor, wherein a signal-to-noise ratio (SNR) associated with the obtained processed signal is greater than or equal to a threshold value; obtaining, by at least one circuit of the plurality of circuits, a reference noise signal from the received second signal based on subtraction of an estimated speech signal present in the received second signal and the obtained processed signal; and determining, by at least one circuit of the plurality of circuits, a processed speech signal based on filtration of the obtained reference noise signal from the obtained processed signal. ### Claim 9 (depends on 8) The method according to claim 8, further comprising: estimating, by at least circuit one of the plurality of circuits, a component of noise corresponding to the at least one noise signal in the received second signal; filtering, by at least one circuit of the plurality of circuits, the estimated component of noise from the received second signal; and obtaining, by at least one circuit of the plurality of circuits, a filtered second signal that comprises at least one of the speech signal or a reverberated speech signal. ### Claim 10 (depends on 8) The method according to claim 8, further comprising: obtaining, by at least one circuit of the plurality of circuits, a single speech signal based on modification of the at least one of the speech signal or the reverberated speech signal in the filtered second signal, wherein the single speech signal is associated with a phase and a gain that matches with corresponding phase and gain of the speech signal in the received first signal; and obtaining, by at least one circuit of the plurality of circuits, the processed signal based on constructive combination of the obtained single speech signal and the received first signal, wherein the processed signal is associated with a gain amplified by the determined factor. ### Claim 11 (depends on 8) The method according to claim 8, further comprising determining, by at least one circuit of the plurality of circuits, a noise path impulse response corresponding to a noise path and a speech path impulse response corresponding to a speech path, wherein the noise path corresponds to a path of reception of the at least one noise signal, and wherein the speech path corresponds to a path of reception of the speech signal. ### Claim 12 (depends on 8) The method according to claim 8, further comprising: determining, by at least one circuit of the plurality of circuits, each formant in a set of formants in the speech signal in the first signal, wherein each formant corresponds to a resonant frequency bin in the speech signal; identifying, by at least one circuit of the plurality of circuits, a defined number of frequency bins of a defined window size adjacent to each determined formant in the set of formants in a specified pattern for adaptation to the speech signal in the received first signal; and estimating, by at least one circuit of the plurality of circuits, an adapted step-size corresponding to each determined formant and each frequency bin of the identified defined number of frequency bins in the speech signal present in the received first signal; and filtering, by at least one circuit of the plurality of circuits, the speech signal from the at least one noise signal based on the estimated step-size. ### Claim 13 A signal processing apparatus, comprising: a first speech transducer configured to receive a first signal from a determined source, wherein the first signal comprises at least one of a speech signal or at least one noise signal; a second speech transducer configured to receive a second signal from the determined source, wherein the second signal comprises at least one of the speech signal or the at least one noise signal; a first pre-processor configured to process the received first signal from the first speech transducer and the second signal from the second speech transducer to obtain a processed signal based on the received first signal and the received second signal for amplification of a gain associated with the speech signal present in the received first signal and the second signal by a determined factor, wherein a signal-to-noise ratio (SNR) associated with the obtained processed signal is greater than or equal to a threshold value; a second pre-processor configured to: estimate the speech signal in the received second signal received from the second speech transducer; and obtain a reference noise signal based on subtraction of the processed signal and the estimated speech signal in the received second signal from the second speech transducer; and a post-processor configured to determine a processed speech signal based on filtration of the reference noise signal from the processed signal. ### Claim 14 (depends on 13) The signal processing apparatus according to claim 13, wherein the first speech transducer is positioned at a first location and the second speech transducer is positioned at a second location, and wherein the first location of the first speech transducer is pre-specified with respect to the second location of the second speech transducer. ### Claim 15 (depends on 13) The signal processing apparatus according to claim 13, wherein the first pre-processor further comprises a first adaptive filter unit, wherein the first adaptive filter unit is configured to: estimate a component of noise corresponding to the at least one noise signal in the second signal received from the second speech transducer; filter the estimated component of noise from the second signal received from the second speech transducer; and obtain a filtered second signal comprising at least one of the speech signal or a reverberated speech signal. ### Claim 16 (depends on 13) The signal processing apparatus according to claim 13, wherein the first pre-processor further comprises a second adaptive filter unit coupled to the first adaptive filter unit, wherein the second adaptive filter unit is configured to: obtain a single speech signal based on modification of the at least one of the speech signal or the reverberated speech signal in the filtered second signal, wherein the single speech signal is associated with a phase and a gain that matches with corresponding phase and gain of the speech signal in the received first signal; and obtain the processed signal based on constructive combination of the obtained single speech signal and the received first signal, wherein the obtained processed signal is associated with a gain amplified by the determined factor. ### Claim 17 (depends on 16) The signal processing apparatus according to claim 16, wherein the second adaptive filter unit is associated with a desired response, wherein the desired response of the second adaptive filter unit is dependent on a noise path and a speech path, and wherein the desired response of the second adaptive filter is inversely proportional to a difference in a speech path impulse response and a noise path impulse response. ### Claim 18 (depends on 16) The signal processing apparatus according to claim 16, wherein the second adaptive filter unit is configured to determine a noise path impulse response corresponding to a noise path and a speech path impulse response corresponding to a speech path, wherein the noise path corresponds to a path of reception of the at least one noise signal from at least one of the first speech transducer or the second speech transducer, and wherein the speech path corresponds to a path of reception of the speech signal from the at least one of the first speech transducer or the second speech transducer. ### Claim 19 (depends on 16) The signal processing apparatus according to claim 16, wherein the second adaptive filter unit is further configured to adapt to filter the speech signal based on reception of the speech signal from the speech path corresponding to the first speech transducer, wherein the second adaptive filter unit is configured to: determine each formant in a set of formants in the speech signal in the first signal received from the first speech transducer, wherein each formant corresponds to a resonant frequency bin in the speech signal; identify a defined number of frequency bins of a defined window size adjacent to each determined formant in the set of formants in a specified pattern for adaptation to the speech signal in the received first signal; and estimate an adapted step-size corresponding to each determined formant and each frequency bin of the identified defined number of frequency bins in the speech signal present in the received first signal; and filter the speech signal from the received first signal based on the estimated step-size. ### Claim 20 (depends on 16) The signal processing apparatus according to claim 16, wherein the second adaptive filter is configured as a sparseness controlled multiline delay filter for determination of each formant and corresponding frequency bins in the speech signal of the first signal. ### Claim 21 (depends on 13) The signal processing apparatus according to claim 13, wherein each of the first speech transducer and the second speech transducer is a microphone, wherein the first pre-processor is an adaptive beamformer, wherein the second pre-processor is a speech remover, and wherein the post-processor is an adaptive noise canceller. ### Claim 22 (depends on 13) The signal processing apparatus according to claim 13, wherein the second pre-processor comprises a first filter and a second filter, wherein the first filter is configured to filter the component of noise in the processed signal received from the first pre-processor, wherein the second filter is configured to filter the estimated the speech signal from the received second signal from the second speech transducer to obtain the reference noise signal, and wherein the second filter receives the filtered first signal from the first filter as a reference signal for the speech signal for removal of the estimated the speech signal from the received second signal. ### Claim 23 (depends on 22) The signal processing apparatus according to claim 22, wherein the first filter in the second pre-processor is a Weiner filter, wherein the second filter in the second pre-processor is a short adaptive filter, and wherein the first filter in the second pre-processor is coupled to the second filter in the second pre-processor. ### Claim 24 (depends on 22) The signal processing apparatus according to claim 22, wherein the second filter in the second pre-processor is adapted for a speech estimation error based on reception of the speech signal from the first speech transducer and the second speech transducer, and wherein the second filter is configured to: determine each formant in a set of formants in the speech signal present in the received second signal from the second speech transducer; and set a step-size and a set of coefficients of the second filter for the adaptation of the second filter to the speech estimation error, wherein the set of coefficients is initialized with zero value based on absence of the speech signal in a speech path of the second speech transducer. ### Claim 25 (depends on 13) The signal processing apparatus according to claim 13, wherein the speech signal from the determined speech source is detected agnostic of a presence of at least one of the speech signal or the at least one noise signal from the first speech transducer and the second speech transducer. ### Claim 26 (depends on 13) The signal processing apparatus according to claim 13, wherein at least the first signal and the second signal comprise at least one silence region, wherein the at least one silence region occurs at least one time interval within the first signal or the second signal.