# US 7,613,532 B2 · Systems and methods for improving the signal to noise ratio for audio input in a computing system > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 7,613,532 B2 - Document ID: US7613532B2 - Status: Granted - Kind: B2 - Application number: 10/704,956 - Filed: 2003-11-10 - Published: 2005-05-12 - Granted: 2009-11-03 - Earliest claimed: 2003-11-10 - Assignees: Microsoft Corporation - Inventors: Jason M. Anderson; Theodore C. Tanner - Web page: https://patentgrep.com/patent/US7613532B2 ## CPC - G06F 3/16: Sound input; Sound output (speech processing G10L) - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) ## Abstract Systems and methods for improving the signal to noise ratio for audio input in a computing system are provided. Signal to noise ratio improvements are enabled for an input audio signal by incorporating a plurality of microphones into the PC environment, for example, to improve the processing of voice or speech input or other audio capture applications. In various embodiments, a plurality of microphones are spaced apart, e.g., attached to or within a PC housing, in such a way that noise from PC components can be “subtracted” or reduced from an input audio signal, which increases the signal to noise ratio and improves audio processing accuracy. The “subtraction” or reduction techniques applied by the present invention are unique to the PC environment wherein different noises having particular characteristics can be identified, including, but not limited to, rattle of device components, fan noise and disk noise. ## Claims ### Claim 1 A method for improving the signal to noise ratio for an audio input signal relative to the noise characteristics of an operating computing device recording the audio input signal, comprising: receiving by the computing device the audio input signal; recording a first sound signal based on the audio input signal with at least one audio input recording device located outside a substantially enclosed housing of the computing device; recording a second sound signal based on the audio input signal with at least one audio input recording device located inside the substantially enclosed housing of the computing device; loading a noise profile stored in the computing device, the noise profile indicative of noise characteristics of the computing device as established at the time of manufacturer of the computing device; generating an estimated noise signal from the noise profile and the second sound signal; and processing the estimated noise signal and the first sound signal to remove the estimated noise signal thereby generating a third sound signal, wherein the signal to noise ratio associated with the third sound signal is higher than the signal to noise ratio of the computing device associated with the first sound signal. ### Claim 2 (depends on 1) A method according to claim 1, wherein the signal to noise ratio associated with the third sound signal is higher than the signal to noise ratio of the computing device associated with the second sound signal. ### Claim 3 (depends on 1) A method according to claim 1, wherein said processing includes processing the first and the estimated noise sound signals to deemphasize at least one of (a) noise from rattle of at least one component in the substantially enclosed housing, (b) fan noise and (c) disk noise. ### Claim 4 (depends on 3) A method according to claim 3, wherein said processing includes processing the first and the estimated noise sound signals to deemphasize disk noise includes deemphasizing noise from at least one of a floppy disk drive, a hard disk and an optical drive. ### Claim 5 (depends on 3) A method according to claim 3, wherein said processing includes processing the first and the estimated noise sound signals to deemphasize disk noise having a periodic and non-stationary noise characteristics. ### Claim 6 (depends on 3) A method according to claim 3, wherein said processing includes processing the first and the estimated noise sound signals to deemphasize fan noise having a noise characteristic that is substantially periodic, and slightly variable. ### Claim 7 (depends on 3) A method according to claim 3, wherein said processing includes processing the first and the estimated noise sound signals to deemphasize noise from rattle including a mechanical disturbance in the housing having a Brownian noise characteristic. ### Claim 8 (depends on 3) A method according to claim 3, wherein said processing includes processing the first and the estimated noise sound signals to deemphasize reflections of at least one of (a) noise from rattle of at least one component in the substantially enclosed housing, (b) fan noise and (c) disk noise. ### Claim 9 (depends on 1) A method according to claim 1, wherein said first and second the estimated noise sound signal are any of analog and digital signals. ### Claim 10 (depends on 1) A method according to claim 1, wherein the processing includes applying variable gain or attenuation to at least one of said first and the estimated noise sound signals. ### Claim 11 (depends on 1) A method according to claim 1, wherein the processing includes one of summing (a) the first and the estimated noise sound signals, (b) the estimated noise sound signal and the output of a variable gain/attenuation component that receives the first sound signal, (c) the first sound signal and the output of a variable gain/attenuation component that receives the estimated noise sound signal and (d) the output of a variable gain/attenuation component that receives the first sound signal and the output of a variable gain/attenuation component that receives the estimated noise sound signal. ### Claim 12 (depends on 1) A method according to claim 1, wherein the processing includes summing the estimated noise sound signal and the output of a variable gain/attenuation component that receives the first sound signal. ### Claim 13 (depends on 1) A method according to claim 1, wherein the processing further includes applying a filter to an intermediate signal based on said first and the estimated noise sound signals, wherein the filter is based upon the noise characteristics of the operating computing device. ### Claim 14 (depends on 12) A method according to claim 12, wherein said applying of the filter includes applying a filter with a stopband rejection characteristic, wherein the frequency range of the stopband rejection can be increased or decreased and the spread of the range from low frequency to high frequency can be changed based on the noise characteristics of the operating computing device. ### Claim 15 (depends on 1) A method according to claim 1, wherein said recording a first sound signal includes recording a first analog sound signal and wherein said recording a second sound signal includes recording a second analog sound signal, the method further comprising: routing the first and second analog sound signals through at least one analog-to-digital converter to generate first and second digital sound signals, respectively. ### Claim 16 (depends on 15) A method according to claim 15, further including: intercepting the first digital sound signal by a frequency response filter and routing the output of the frequency response filter to a digital signal processing (DSP) device. ### Claim 17 (depends on 16) A method according to claim 16, wherein said processing includes processing the output of the frequency response filter and the estimated noise digital sound signal with computer-executable instructions stored on a computer-executable medium in the operating computing device. ### Claim 18 (depends on 17) A method according to claim 17, wherein said processing includes applying variable gain/attenuation, summing and a transfer function to at least one signal based on the output of the frequency response filter and the estimated noise digital sound signal. ### Claim 19 (depends on 1) A method according to claim 1, further including outputting the third sound signal to a component of the computing device utilized in connection with at least one of a speech recognition application, a speech activated application, a real-time communication (RTC) application, a speech to text application, a voice over internet protocol (VOIP) application, a music recognition application. ### Claim 20 (depends on 1) A method according to claim 1, further including outputting the third sound signal to an I/O controller, which forwards the third sound signal to a sound card. ### Claim 21 (depends on 1) A method according to claim 1, further including encoding the third sound signal according to a pre-set encoding/decoding algorithm. ### Claim 22 (depends on 21) A method according to claim 21, further including decoding the third sound signal according to the pre-set encoding/decoding algorithm. ### Claim 23 (depends on 1) A method according to claim 1, wherein said recording the second sound signal based on the audio input signal with at least one audio input recording device located inside the substantially enclosed housing of the computing device includes recording the second sound signal with a plurality of microphones spaced apart inside the substantially enclosed housing. ### Claim 24 (depends on 1) A method according to claim 1, wherein the noise characteristics of the operating computing device substantially originate from inside the substantially enclosed housing of the operating computing device, with ambient noise generally being shielded by the housing. ### Claim 25 (depends on 1) A method according to claim 1, wherein the processing includes comparing the first sound signal and the second sound signal and amplifying the common frequencies. ### Claim 26 (depends on 1) A method according to claim 1, wherein the recording the first sound signal based on the audio input signal with at least one audio input recording device located outside a substantially enclosed housing of the computing device includes recording the first sound signal with at least one microphone and at least one array microphone. ### Claim 27 (depends on 1) A method according to claim 1, wherein at least one of (A) the recording the first sound signal based on the audio input signal with at least one audio input recording device and (B) the recording the second sound signal based on the audio input signal with at least one audio input recording device includes characterizing the at least one audio input recording device in advance with a frequency response filter for improved recording. ### Claim 28 (depends on 1) A method according to claim 1, wherein said processing includes dynamically adjusting the processing according to a filter that adjusts to the proximity of the user to the at least one recording device of the recording of the first sound signal. ### Claim 29 (depends on 1) A method according to claim 1, wherein said processing includes filtering to adjust for changes in the noise characteristics of the computing device over time to maintain optimum recording. ### Claim 30 (depends on 1) A method according to claim 1, wherein said processing includes processing in the frequency domain. ### Claim 31 A computing device, including: a plurality of physical components including a substantially enclosed housing; at least one audio input recording device located outside the substantially enclosed housing for recording an audio signal from outside the substantially enclosed housing during a time period and generating a first sound signal based thereon; at least one audio input recording device located inside the substantially enclosed housing for recording the audio signal from inside the substantially enclosed housing during the time period and generating a second sound signal based thereon, wherein for the first sound signal, the ratio of the audio signal to a noise signal arising from noise characteristics of at least one physical component of the plurality of physical components is greater than the ratio of the audio signal to the noise signal for the second sound signal; and at least one signal processing component configured to load a noise profile stored in the computing device, the noise profile indicative of noise characteristics of the computing device as established at the time of manufacturer of the computing device, the signal processing component further configured to generate an estimated noise signal from the noise profile and the second sound signal and the computing device configured to process the estimated noise signal and the first sound signal to remove the estimated noise signal thereby generating a third sound signal representative of the audio signal, wherein ratio of the audio signal to noise signal associated with the third sound signal is higher than the ratios for both the first and second sound signals. ### Claim 32 (depends on 31) A computing device according to claim 31, wherein said at least one signal processing component is configured to remove at least one of (a) noise from rattle of at least one component in the substantially enclosed housing, (b) fan noise and (c) disk noise. ### Claim 33 (depends on 32) A computing device according to claim 32, wherein said at least one signal processing component is configured to remove reflections of the at least one of (a) noise from rattle of at least one component in the substantially enclosed housing, (b) fan noise and (c) disk noise. ### Claim 34 (depends on 31) A computing device according to claim 31, wherein the at least one signal processing component includes a variable gain or attenuation component that receives as input at least one of said first and second sound signals. ### Claim 35 (depends on 31) A computing device according to claim 31, wherein the at least one signal processing component includes a summing component that receives as input the second sound signal and the output of a variable gain/attenuation component that receives the first sound signal. ### Claim 36 (depends on 31) A computing device according to claim 31, wherein the at least one signal processing component includes a filter component for filtering based upon the noise characteristics of the noise signal. ### Claim 37 (depends on 31) A computing device according to claim 31, wherein the at least one signal processing component includes at least one analog-to-digital converter to generate first and second digital sound signals from said first and second sound signals, respectively. ### Claim 38 (depends on 37) A computing device according to claim 37, wherein the at least one signal processing component includes a frequency response filter which receives the first sound signal. ### Claim 39 (depends on 38) A computing device according to claim 38, wherein the at least one signal processing component includes computer-executable instructions stored on a computer-executable medium in the operating computing device for applying variable gain/attenuation, summing and a transfer function to at least one signal based on the output of the frequency response filter and the second digital sound signal. ### Claim 40 (depends on 31) A computing device according to claim 31, further including a sound card that receives the third sound signal as input. ### Claim 41 (depends on 31) A computing device according to claim 31, wherein the at least one signal processing component includes an encoder/decoder (codec) component. ### Claim 42 (depends on 31) A computing device according to claim 31, wherein the at least one signal processing component compares the first sound signal and the estimated noise signal and amplifies the common frequencies. ### Claim 43 (depends on 31) A computing device according to claim 31, wherein the at least one signal processing component includes a dynamic filter that adjusts to the proximity of the user to the at least one recording device of the recording of the first sound signal. ### Claim 44 (depends on 31) A computing device according to claim 31, wherein the at least one signal processing component includes a filter component that adjusts for changes in the noise signal of the computing device over time to maintain optimum recording. ### Claim 45 (depends on 31) A computing device according to claim 31, wherein the at least one signal processing component includes a filter that filters according to a noise globally unique identifier (GUID) assigned to the plurality of components of the computing device. ### Claim 46 (depends on 31) A computing device according to claim 31, wherein said at least one signal processing component processes the first and second sound signals in the frequency domain. ### Claim 47 Signal processing electronics for processing sound signals in a computing device containing a plurality of physical components including a substantially enclosed housing, the signal processing electronics receiving as input a first sound signal recorded from outside the substantially enclosed housing of the computing device and a second sound signal recorded from inside the substantially enclosed housing of the computing device, comprising: circuitry for loading a noise profile stored in the computing device, the noise profile indicative of noise characteristics of the computing device as established at the time of manufacturer of the computing device; circuitry for generating an estimated noise signal from the noise profile and the second sound signal; and circuitry for processing the estimated noise signal and the first sound signal to remove the estimated noise signal thereby generating a third sound signal, wherein the signal to noise ratio associated with the third sound signal is higher than the signal to noise ratio of the computing device associated with the first sound signal. ### Claim 48 (depends on 47) Signal processing electronics according to claim 47, further including at least one analog-to-digital converter to generate first and second digital sound signals from said first and second sound signals, respectively. ### Claim 49 (depends on 48) Signal processing electronics according to claim 48, further including a frequency response filter which receives the first sound signal. ### Claim 50 (depends on 47) Signal processing electronics according to claim 47, further including an encoder/decoder (codec) component. ### Claim 51 (depends on 47) Signal processing electronics according to claim 47, wherein said circuitry for processing the estimated noise signal and the first sound signal is configured to compare the first sound signal and the estimated noise sound signal and amplify the common frequencies. ### Claim 52 (depends on 47) Signal processing electronics according to claim 47, wherein said circuitry for processing the estimated noise signal and the first sound signal is configured to adjust for changes in noise characteristics of a noise signal created by the computing device over time. ### Claim 53 (depends on 47) Signal processing electronics according to claim 47, wherein said circuitry for loading the noise profile stored in the computing device is configured to load the noise profile filter component filters according to a noise globally unique identifier (GUID) assigned to a plurality of components of the computing device. ### Claim 54 (depends on 47) Signal processing electronics according to claim 47, wherein said circuitry for processing the estimated noise signal and the first sound signal is configured to remove at least one of (a) noise from rattle of at least one component in the substantially enclosed housing, (b) fan noise and (c) disk noise. ### Claim 55 (depends on 54) Signal processing electronics according to claim 54, wherein said circuitry for processing the estimated noise signal and the first sound signal is configured to remove reflections of the at least one of (a) noise from rattle of at least one component in the substantially enclosed housing, (b) fan noise and (c) disk noise. ### Claim 56 (depends on 54) Signal processing electronics according to claim 54, wherein said signal processing electronics process sounds signals in the frequency domain. ### Claim 57 A processor operatively coupled to a computing readable storage medium including computer readable instructions for improving the signal to noise ratio for an audio input signal relative to the noise characteristics of an operating computing device recording the audio input signal, when executed, the computer readable storage medium carry out the following methods: recording a first sound signal based on the audio input signal with at least one audio input recording device located outside a substantially enclosed housing of the computing device; recording a second sound signal based on the audio input signal with at least one audio input recording device located inside the substantially enclosed housing of the computing device; loading a noise profile stored in the computing device, the noise profile indicative of noise characteristics of the computing device as established at the time of manufacturer of the computing device; generating an estimated noise signal from the noise profile and the second sound signal; and processing the estimated noise signal and the first sound signal to remove the estimated noise signal thereby generating a third sound signal, wherein the signal to noise ratio associated with the third sound signal is higher than the signal to noise ratio of the computing device associated with the first sound signal. ### Claim 58 (depends on 57) The processor operatively coupled to a computer readable storage medium of claim 57, wherein processing the estimated noise signal and the first sound signal further comprise processing the first and second sound signals in the frequency domain.