# US 10,313,782 B2 · Automatic speech recognition triggering system > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 10,313,782 B2 - Document ID: US10313782B2 - Status: Granted - Kind: B2 - Application number: 15/587,325 - Filed: 2017-05-04 - Published: 2018-11-08 - Granted: 2019-06-04 - Earliest claimed: 2017-05-04 - Assignees: Apple Inc. - Applicants: Apple Inc. - Inventors: Sorin V. Dusan; Aram M. Lindahl; Robert D. Watson - Web page: https://patentgrep.com/patent/US10313782B2 ## CPC - H04R 1/1091: {Details not provided for in groups H04R1/1008 - H04R1/1083} - G06F 3/167: {Audio in a user interface, e.g. using voice commands for navigating, audio feedback} - G10L 15/08: Speech classification or search - G10L 25/06: the extracted parameters being correlation coefficients - G10L 25/78: Detection of presence or absence of voice signals (switching of direction of transmission by voice frequency in two-way loud-speaking telephone systems H04M9/10) - G10L 15/22: Procedures used during a speech recognition process, e.g. man-machine dialogue - G10L 2015/088: {Word spotting} - H04R 2460/13: Hearing devices using bone conduction transducers - H04R: LOUDSPEAKERS, MICROPHONES, GRAMOPHONE PICK-UPS OR LIKE ACOUSTIC ELECTROMECHANICAL TRANSDUCERS; ELECTRIC HEARING AIDS; PUBLIC ADDRESS SYSTEMS (producing sounds with frequency not determined by supply frequency G10K) - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) - G10L: SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING ## Abstract An automatic speech recognition (ASR) triggering system, and a method of providing an ASR trigger signal, is described. The ASR triggering system can include a microphone to generate an acoustic signal representing an acoustic vibration and an accelerometer worn in an ear canal of a user to generate a non-acoustic signal representing a bone conduction vibration. A processor of the ASR triggering system can receive an acoustic trigger signal based on the acoustic signal and a non-acoustic trigger signal based on the non-acoustic signal, and combine the trigger signals to gate an ASR trigger signal. For example, the ASR trigger signal may be provided to an ASR server only when the trigger signals are simultaneously asserted. Other embodiments are also described and claimed. ## Claims ### Claim 1 An automatic speech recognition (ASR) triggering system, comprising: a microphone to generate an acoustic signal representing an acoustic vibration; an accelerometer to generate a non-acoustic signal representing a bone conduction vibration; an acoustic detector to receive the acoustic signal and to generate an acoustic trigger signal in response to the acoustic signal matching an energy signature of a key-phrase; a voice activity detector (VAD) to receive the non-acoustic signal and to generate a VAD signal based on the non-acoustic signal; a pattern detector to receive the VAD signal and to generate a non-acoustic trigger signal in response to the VAD signal matching a predetermined sequence of energy intervals of the key phrase; and a processor to receive the acoustic trigger signal and the non-acoustic trigger signal, and generate an ASR trigger signal based on a combination of the acoustic trigger signal and the non-acoustic trigger signal when the acoustic trigger signal and the non-acoustic trigger signal are asserted. ### Claim 2 (depends on 1) The ASR triggering system of claim 1, wherein the accelerometer detects the bone conduction vibration along one or more axes, wherein the non-acoustic signal includes one or more axis signals, and wherein the VAD generates the VAD signal based on an energy in the one or more axis signals. ### Claim 3 (depends on 2) The ASR triggering system of claim 2, wherein the one or more axis signals includes a first axis signal and a second axis signal, and wherein the VAD generates the VAD signal based on a cross-correlation of the first axis signal and the second axis signal. ### Claim 4 (depends on 1) The ASR triggering system of claim 1, wherein the VAD receives the acoustic signal, and wherein the VAD cross-correlates the acoustic signal and the non-acoustic signal to generate the VAD signal. ### Claim 5 (depends on 1) The ASR triggering system of claim 1, wherein the pattern detector detects a match between the VAD signal and the predetermined sequence of energy intervals, and generates the non-acoustic trigger signal in response to the match. ### Claim 6 (depends on 1) The ASR triggering system of claim 1, wherein the acoustic detector generates the acoustic trigger signal as a high digital signal, and wherein the VAD generates the VAD signal as the high digital signal when an energy or a cross-correlation value of the non-acoustic signal is above a predetermined threshold. ### Claim 7 (depends on 6) The ASR triggering system of claim 6, wherein the processor generates the ASR trigger signal when the acoustic trigger signal and the non-acoustic trigger signal are simultaneously high digital signals. ### Claim 8 (depends on 1) The ASR triggering system of claim 1 further comprising a second acoustic detector to receive the acoustic signal and to generate a power-on signal in response to detecting a power-on portion of the energy signature of the key-phrase, and wherein the accelerometer powers on and generates the non-acoustic signal in response to receiving the power-on signal from the second acoustic detector. ### Claim 9 (depends on 1) The ASR triggering system of claim 1, wherein the VAD receives the acoustic signal to generate the VAD signal based on the acoustic signal and the non-acoustic signal. ### Claim 10 An automatic speech recognition (ASR) triggering system, comprising: a microphone to generate an acoustic signal representing an acoustic vibration corresponding to user vocalizations, wherein the user vocalizations include a key-phrase and one or more additional vocalizations other than the key-phrase; an accelerometer to generate a non-acoustic signal representing a bone conduction vibration corresponding to the user vocalizations; and a processor including an acoustic channel to receive the acoustic signal and a non-acoustic channel to receive the non-acoustic signal, wherein the processor generates an ASR trigger signal based on a combination of the acoustic signal and the non-acoustic signal, and wherein the processor generates the ASR trigger signal to trigger speech recognition functions on the one or more additional vocalizations other than the key-phrase at least in response to the acoustic signal and the non-acoustic signal match a sequence of energy intervals of the key-phrase. ### Claim 11 (depends on 10) The ASR triggering system of claim 10, wherein the processor generates the ASR trigger signal when the acoustic signal matches the non-acoustic signal. ### Claim 12 (depends on 11) The ASR triggering system of claim 11, wherein the acoustic signal matches the non-acoustic signal when the acoustic signal and the non-acoustic signal simultaneously match the sequence of energy intervals of the key-phrase. ### Claim 13 (depends on 10) The ASR triggering system of claim 10, wherein the acoustic signal has a higher energy bandwidth than the non-acoustic signal. ### Claim 14 (depends on 10) The ASR triggering system of claim 10, wherein the ASR trigger signal is a binary output. ### Claim 15 An automatic speech recognition (ASR) triggering system, comprising: a microphone to generate an acoustic signal corresponding to user vocalizations, wherein the user vocalizations include a sequence of hums and one or more additional vocalizations other than the sequence of hums; an accelerometer to generate a non-acoustic signal corresponding to the sequence of hums made by a user, wherein the sequence of hums have respective durations; and a processor to receive the non-acoustic signal, and generate an ASR trigger signal based on a comparison of the respective durations of the sequence of hums to a selected sequence of durations of a hum pattern, wherein the ASR trigger signal is generated by the processor to trigger speech recognition functions on the one or more additional vocalizations other than the sequence of hums at least in response to the respective durations match the selected sequence of durations. ### Claim 16 (depends on 15) The ASR triggering system of claim 15, wherein the processor determines a sequence of energy peaks of the non-acoustic signal corresponding to the sequence of hums, compares the sequence of energy peaks to a selected sequence of energy intervals having the selected sequence of durations, and generates the ASR trigger signal in response to determining the sequence of energy peaks matches the selected sequence of energy intervals. ### Claim 17 (depends on 16) The ASR triggering system of claim 16, wherein the accelerometer detects a bone conduction vibration along a first axis and along a second axis, wherein the non-acoustic signal includes a first axis signal corresponding to the bone conduction vibration along the first axis and a second axis signal corresponding to the bone conduction vibration along the second axis, and wherein the processor cross-correlates the first axis signal and the second axis signal to generate a VAD signal having the sequence of energy peaks. ### Claim 18 (depends on 17) The ASR triggering system of claim 17, wherein the processor determines an energy peak of the energy peaks when a cross-correlation value of the first axis signal and the second axis signal is above a selected correlation threshold. ### Claim 19 (depends on 18) The ASR triggering system of claim 18, wherein each respective duration of the sequence of hums is between when the cross-correlation value rises above the selected correlation threshold to when the cross-correlation value falls below the selected correlation threshold.