Speech recognition
A speech recognition system comprises: an input, for receiving an input signal from at least one microphone; a first buffer, for storing the input signal; a noise reduction block, for receiving the input signal and generating a noise reduced input signal; a speech recognition engine, for receiving either the input signal output from the first buffer or the noise reduced input signal from the noise reduction block; and a selection circuit for directing either the input signal output from the first buffer or the noise reduced input signal from the noise reduction block to the speech recognition engine.
1. A method of processing a signal representing speech, the method comprising:
receiving signals from a plurality of microphones;
selecting a signal from one of said plurality of microphones;
continuously buffering the selected signal, for a first period of time longer than an expected duration of a predetermined pass phrase;
while continuously buffering the selected signal, monitoring the selected signal to detect the presence of speech; and
on detecting the presence of speech in the received selected signal:
continuously buffering the selected signal for a second period of time,
wherein the second period of time is longer than the first period of time; and
attempting to detect the predetermined pass phrase in the content of the buffer.
2. The method according to claim 1 , comprising buffering the selected signal for a period of time in the range of 2-10 seconds.
3. The method according to claim 1 , comprising continuously buffering the selected signal on a first-in first-out basis.
4. The method according to claim 1 , comprising:
receiving analog signals from said plurality of microphones;
converting the selected signal into digital form; and
continuously buffering the selected signal in digital form.
5. The method according to claim 1 , when performed in a mobile phone, the method comprising:
when the signals from the plurality of microphones comprise a signal from a headset having a connection to the mobile phone, selecting said signal from the headset.
6. The method according to claim 1 , comprising:
determining whether at least one of said plurality of microphones is occluded;
wherein the selected signal is from one of said plurality of microphones that has been determined not to be occluded.
7. The method according to claim 1 , further comprising:
on detecting the predetermined pass phrase in the content of the buffer, performing speaker recognition to determine whether the predetermined pass phrase was spoken by an enrolled speaker.
8. A signal processing device, comprising:
an input for receiving signals from a plurality of microphones;
a selector for selecting a signal from one of said plurality of microphones;
a buffer for continuously buffering the selected signal, for a period of time longer than an expected duration of a predetermined pass phrase;
a signal activity detector, for monitoring the selected signal during the continuous buffering of the selected signal to detect the presence of speech; and
a pass phrase detector, for attempting to detect the predetermined pass phrase in the content of the buffer following detection of the presence of speech in the selected signal;
wherein the buffer is configured for:
continuously buffering the selected signal, for a first period of time; and
in response to detecting the presence of speech in the selected signal, continuously buffering the selected signal for a second period of time, wherein the second period of time is longer than the first period of time.
9. The device according to claim 8 , wherein the buffer is capable of storing the selected signal for a period of time in the range of 2 -10 seconds.
10. The device according to claim 8 , wherein the buffer is configured for continuously storing the selected signal on a first-in first-out basis.
11. The device according to claim 8 , further comprising:
a speaker recognition block, for performing speaker recognition to determine whether the predetermined pass phrase was spoken by an enrolled speaker, following detection of the predetermined pass phrase in the content of the buffer, to determine whether the predetermined pass phrase was spoken by an enrolled speaker.
12. A mobile phone, comprising a plurality of microphones, and a signal processing device, wherein the signal processing device comprises:
an input for receiving signals from the plurality of microphones;
a selector for selecting a signal from one of the plurality of microphones;
a buffer for continuously buffering the selected signal, for a period of time longer than an expected duration of a predetermined pass phrase;
a signal activity detector, for monitoring the selected signal to detect the presence of speech during the continuous buffering of the selected signal; and
a pass phrase detector, for attempting to detect the predetermined pass phrase in the content of the buffer following detection of the presence of speech in the selected signal;
wherein the buffer is configured for:
continuously buffering the selected signal, for a first period of time; and
in response to detecting the presence of speech in the selected signal, continuously buffering the selected signal for a second period of time, wherein the second period of time is longer than the first period of time.
13. The mobile phone according to claim 12 ,
wherein the plurality of microphones are configured for generating analog signals;
comprising an analog-digital converter for converting the selected signal into digital form;
wherein the buffer is configured for continuously buffering the selected signal in digital form.
14. The mobile phone according to claim 12 , having a connection for a headset, wherein the selector is configured for, when a headset is connected thereto, selecting said signal from the headset.
15. The mobile phone according to claim 12 , configured for:
determining whether at least one of said plurality of microphones is occluded;
wherein the selected signals from one of said plurality of microphones that has been determined not to be occluded.