IP Library Granted Patent US 11,087,780
Granted Patent B2
US 11,087,780 · App. 16/228,351 · Granted Aug 10, 2021

Analog voice activity detector systems and methods

Inventors: Lorenzo Crespi (Costa Mesa, CA); Marco Croce (Orio Litta, IT); Andrea Baschirotto (Tortona, IT); Piero Malcovati (Pavia, IT)
Assignee: SYNAPTICS INCORPORATED
G10L25/84G10L15/22G10L25/21G10L2015/223
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 11,087,780
App. No.
16/228,351
Granted
Aug 10, 2021
Kind
B2
Abstract

A voice-activity detector (VAD) system comprises a microphone operable to receive and to process audio inputs from an environment to generate an analog audio input signal. The system further comprises an analog VAD operable to process the analog audio input signal to perform an initial detection of human speech, and operable to send a wake up command to a digital signal processing chain to awaken the digital signal processing chain from a sleep mode, when the analog VAD detects human speech. Further, the system comprises the digital signal processing chain operable to process the analog audio input signal to perform a secondary detection of human speech, and operable to output a signal indicating that human speech is detected, when the digital signal processing chain detects human speech. In one or more embodiments, the secondary detection of human speech is more accurate than the initial detection of human speech.

Claims (39)

1. A method comprising:

receiving an analog audio input signal;

processing, by an analog voice activity detector, the analog audio input signal to determine whether the analog audio input signal includes speech;

generating, by the analog voice activity detector, a wake up signal to a digital signal processing chain if the analog audio input signal is determined to include speech, the wake up signal notifying the digital signal processing chain to wake from a sleep mode;

converting, by an analog to digital converter of the digital signal processing chain, the analog audio input signal to a digital audio input signal;

detecting speech in the digital audio input signal using a digital signal processor of the digital signal processing chain;

outputting, by the digital signal processing chain if the digital signal processing chain detects speech, a command indicating that human speech has been detected; and

wherein the processing, by the analog voice activity detector, comprises comparing, by analog noise level estimation circuitry comprising a switched capacitance network and a comparator, an energy level of the audio input signal and a noise floor level of the analog audio input signal to a variable signal processing threshold to determine whether the analog audio input signal includes speech, wherein the switched capacitance network comprises a plurality of capacitors and switches, and wherein the switches are operable to adjust the noise floor level based on a comparison of the noise floor level to the energy level.

2. The method of claim 1 , wherein the processing, by the analog voice activity detector, further comprises band limiting and amplifying, by a low-power bandpass pre-amplifier, the analog audio input signal to produce a band-limited amplified analog audio signal.

3. The method of claim 2 , wherein the low-power bandpass pre-amplifier is one of an active resistor-capacitor (RC) circuit, a gm-C circuit, and a passive RC circuit.

4. The method of claim 3 , wherein the processing, by the analog voice activity detector, further comprises

calculating, by an energy estimator, an energy level of the band-limited amplified analog audio signal to produce a voltage sample indicative of a corresponding energy level.

5. The method of claim 4 , wherein the calculating, by the energy estimator, the energy level comprises performing a continuous-time nonlinear operation on the band-limited amplified analog audio signal.

6. The method of claim 5 , wherein the processing, by the analog voice activity detector, further comprises:

calculating, by the analog noise level estimation circuitry, an estimation of the noise floor level; and

calculating a signal-to-noise (SNR) level of the voltage sample and the noise floor level.

7. The method of claim 1 , wherein the comparing includes an estimation of a signal to noise ratio of a sample voltage indicative of the energy level.

8. A system comprising:

an audio input operable to sense sound waves from an environment and generate an analog audio input signal;

a digital signal processing chain operable to process the analog audio input signal to detect speech and output a signal indicating when speech is detected, the digital signal processing chain operable to enter a low power mode after a period of inactivity; and

analog voice activity detection circuitry coupled to the audio input and digital signal processing chain, the analog voice activity detection circuitry operable to process the analog audio input signal to detect speech and generate a wake up command to the digital signal processing chain to awaken the digital signal processing chain from the low power mode when speech is detected by the analog voice activity detection circuitry;

wherein the analog voice activity detection circuitry comprises analog noise level estimation circuitry operable to calculate an estimation of a noise floor level, the analog noise level estimation circuitry comprising a switched capacitance network and a comparator; and analog signal-to-noise level circuitry operable to compare a signal-to-noise ratio to a variable threshold value.

9. The system of claim 8 wherein the analog voice activity detection circuitry further comprises:

analog energy estimation circuitry operable to receive the analog input audio signal and generate an energy level estimation for a frame of the input audio signal.

10. The system of claim 9 , wherein the analog noise level estimation circuitry is coupled to the output of the analog energy estimation circuitry, and wherein the analog noise level estimation circuitry is operable to receive the energy level estimation from the analog energy estimation circuitry and generate an estimated noise level.

11. The system of claim 9 , wherein the analog voice activity detection circuitry further comprises voice activity detection logic circuitry coupled to the output of the energy estimation circuitry and the noise level estimation circuitry, the voice activity detection logic comprising analog circuitry operable to receive the energy estimation from the energy estimation circuit and the signal to noise ratio output from the noise level estimation circuitry and generate a signal indicating whether the speech has been detected.

12. The system of claim 8 , wherein the analog voice activity detection circuitry further comprises a low-power bandpass pre-amplifier operable to output a band limited, amplified analog audio input signal.

13. The system of claim 12 , wherein the low-power bandpass pre-amplifier comprises one of an active resistor-capacitor (RC) circuit, a gm-C circuit, and a passive RC circuit.

14. The system of claim 13 , wherein the analog voice activity detection circuitry further comprises analog energy estimation circuitry operable to receive the analog audio input signal and calculate an estimated energy level to produce a voltage sample indicative of a corresponding energy level.

15. A system comprising:

an audio input operable to sense sound waves from an environment and generate an analog audio input signal;

a digital signal processing chain operable to process the analog audio input signal to detect speech and output a signal indicating when speech is detected, the digital signal processing chain operable to enter a low power mode after a period of inactivity; and

analog voice activity detection circuitry coupled to the audio input and digital signal processing chain, the analog voice activity detection circuitry operable to process the analog audio input signal to detect speech and generate a wake up command to the digital signal processing chain to awaken the digital signal processing chain from the low power mode when speech is detected by the analog voice activity detection circuitry;

wherein the analog voice activity detection circuitry comprises analog energy estimation circuitry operable to receive the analog audio input signal and calculate an estimated energy level to produce a voltage sample indicative of a corresponding energy level; and

wherein the analog energy estimation circuitry is operable to calculate the energy level by performing a continuous-time nonlinear operation on a band-limited amplified analog audio signal, and wherein the analog voice activity detection circuitry is operable to compare the calculated energy level and a calculated noise floor level of the analog audio input signal to a variable signal processing threshold to determine whether the analog audio input signal includes speech.

16. The system of claim 15 , wherein the analog voice activity detection circuitry further comprises:

analog noise level estimation circuitry operable to calculate an estimation of the noise floor level, the analog noise level estimation circuitry comprising a switched capacitance network and a comparator; and

analog signal-to-noise level circuitry operable to compare a signal-to-noise ratio to a variable threshold value.

17. The system of claim 16 , wherein the switched capacitance network comprises a plurality of capacitors and a plurality of switches, and wherein the switches are operable to adjust the noise floor level based on the comparison of the noise floor level to the calculated energy level.

Assignments (2)
SECURITY INTEREST Recorded Feb 14, 2020
From: SYNAPTICS INCORPORATED
To: WELLS FARGO BANK, NATIONAL ASSOCIATION
Reel/Frame 051936/0103 →
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Dec 27, 2018
From: CRESPI, LORENZO; CROCE, MARCO; BASCHIROTTO, ANDREA; MALCOVATI, PIERO
To: SYNAPTICS INCORPORATED
Reel/Frame 047857/0758 →