Adaptive environmental context sample and update for comparing speech recognition

• US 9,412,373 B2
• Filed: 08/28/2013
• Issued: 08/09/2016
• Est. Priority Date: 08/28/2013
• Status: Active Grant

• Current Assignee: Texas Instruments, Inc.
• Original Assignee: Texas Instruments, Inc.
• Inventors: Ma, Wei, Tan, Bozhao
• Primary Examiner(s): Chawan, Vijay B
• Application Number:
  US14/013,015
• Publication Number:
  US 20150371638A1
• Time in Patent Office: 1,077 Days
• Field of Search: 704/275, 704/255, 704/248, 704/243, 704/250, 704/253, 704/246, 704/233, 704/200.1, 704/216, 704/258, 704/231, 340/539.26, 340/575, 379/1.02, 379/27.03, 381/120, 381/122, 381/303, 381/71.11, 381/92, 381/96, 381/97, 709/224
• US Class Current: 1/1
• CPC Class Codes: G10L 15/02 Feature extraction for speech recognition; Selection of recognition unit
• CPC Class Codes: G10L 15/065 Adaptation
• CPC Class Codes: G10L 15/28 Constructional details of speech recognition systems
• CPC Class Codes: G10L 17/02 Preprocessing operations, e.g. segment selection; Pattern representation or modelling, e.g. based on linear discriminant analysis [LDA] or principal components; Feature selection or extraction
• CPC Class Codes: G10L 2015/228 of application context
• CPC Class Codes: 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

Abstract

A low power sound recognition sensor is configured to receive an analog signal that may contain a signature sound. Sparse sound parameter information is extracted from the analog signal. The extracted sound parameter information is sampled in a periodic manner and a context value is updated to indicate a current environmental condition. The sparse sound parameter information is compared to both the context value and a signature sound parameter database stored locally with the sound recognition sensor to identify sounds or speech contained in the analog signal, such that identification of sound or speech is adaptive to the current environmental condition.

Claims

1. A method for operating a sound recognition sensor, the method comprising:

receiving an analog signal that may contain a signature sound;

extracting sparse sound parameter information from the analog signal;

sampling the extracted sparse sound parameter information in a periodic manner and updating an environmental context value to indicate a current environmental condition; and

comparing the sparse sound parameter information to both the environmental context value and a signature sound parameter database stored locally with the sound recognition sensor to identify sounds or speech contained in the analog signal, such that identification of sound or speech is adaptive to the current environmental condition.

2. The method of claim 1, further comprising adjusting a sound detection threshold in response to the context value.

3. The method of claim 1, further comprising:

placing a digital classification portion of the sound recognition sensor in a sleep state;

comparing an initial truncated portion of the sound parameter information to a truncated sound parameter database stored locally with the sound recognition sensor to detect when there is a likelihood that the expected sound is being received in the analog signal;

generating a trigger signal to awaken the classification portion when the likelihood that the expected sound is being received exceeds a threshold value;

waking the digital classification portion from the sleep state in response to the trigger signal; and

using the digital processing portion for comparing the sound parameter information to both the context value and a signature sound parameter database.

4. The method of claim 3, further comprising adjusting the trigger signal detection based on the context value to reduce the wake frequency of the digital classification portion.

5. The method of claim 1, wherein extracting sparse sound parameter information from the analog signal is performed using an analog portion of the sound recognition sensor.

6. The method of claim 1, wherein extracting the sparse sound parameters is performed at a sample rate of less than or equal to approximately 500 samples per second.

7. The method of claim 6, wherein the sample rate is approximately 50 samples per second for voice sound.

8. An apparatus for recognizing a sound, the apparatus comprising:

a microphone;

an analog front end section comprising analog feature extraction circuitry configured to receive an analog signal from the microphone that may contain a signature sound and to extract sparse sound parameter information from the analog signal;

a digital classification section coupled to the analog front end section;

an environmental context value storage coupled to the digital classification section, wherein the digital classification section is operable to sample the extracted sparse sound parameter information in a periodic manner and to update the environmental context value storage to indicate a current environmental condition; and

wherein the digital classification is configured to compare the sparse sound parameter information to both the environmental context value and a signature sound parameter database stored locally with the digital classification section to identify sounds or speech contained in the analog signal, such that identification of sound or speech is adaptive to the current environmental condition.

9. The apparatus of claim 8, wherein the digital classification section is operable to adjust a sound detection threshold in response to the context value.

10. The apparatus of claim 8, wherein the digital classification section is operable to be placed in a low power sleep state, further comprising:

event trigger circuitry coupled to the analog front end section, wherein the event trigger circuitry is operable to;

compare an initial truncated portion of the sound parameter information to a truncated sound parameter database stored locally with the event trigger circuitry to detect when there is a likelihood that the expected sound is being received in the analog signal; and

wake the digital classification portion from the sleep state in response to detecting a likelihood that the expected sound is being received in the analog signal.

11. The apparatus of claim 8, wherein the analog front end is configured to extract the sparse sound parameters at a sample rate of less than or equal to approximately 500 samples per second.

12. The apparatus of claim 11, wherein the sample rate is approximately 50 samples per second.