Method and system for detecting a relevant utterance in a voice session
View Patent ↗A method and apparatus for detecting use of an utterance. A voice session including voice signals generated during a conversation between a first participant and a second participant is monitored by a speech analytics processor. The speech analytics processor detects the use of an utterance. A speech recognition processor channel selected from a pool of speech recognition processor channels and is coupled to the voice session. The speech recognition processor provided speech recognition services to a voice-enabled application. The speech recognition processor channel is then decoupled from the voice session. The speech analytics processor continues to monitor the conversation for subsequent use of the utterance.
1. A method for monitoring a conversation comprising:
coupling a voice session comprising voice signals generated during a conversation between a first participant and a second participant to a speech analytics processor to detect in the voice signals a use of an utterance requesting a service from a voice enabled application;
receiving from the speech analytics processor an indication that the utterance has been detected in the voice signals and decoupling the speech analytics processor from the voice session;
coupling the voice session to a speech recognition processor which receives a grammar comprising a plurality of commands for the voice enabled application, wherein the speech recognition processor monitors the voice session subsequent to the use of the utterance for one of the plurality of commands; and
receiving from the speech recognition processor an indication that the one of the plurality of commands has been detected and re-coupling the speech analytics processor to the voice session.
2. The method of claim 1 wherein coupling the speech analytics processor to detect in the voice signals the use of the utterance further comprises generating, by the speech analytics processor, a phonetic index comprising phonemes associated with the voice signals, and determining if one or more phonemes in the phonetic index matches one or more phonemes associated with the utterance.
3. The method of claim 1 wherein coupling the voice session to the speech recognition processor further comprises coupling the voice session to the speech recognition processor after the utterance has been detected.
4. The method of claim 3 further comprising decoupling the speech recognition processor from the voice session.
5. The method of claim 3 wherein coupling the voice session to the speech recognition processor comprises coupling the voice session to one of a plurality of speech recognition channels associated with a pool of speech recognition channels, and wherein the one of the plurality of speech recognition channels is selected from the pool of speech recognition channels upon detecting the use of the utterance, and the one of the plurality of speech recognition channels is returned to the pool of speech recognition channels upon decoupling the one of the plurality of speech recognition channels from the voice session.
6. The method of claim 1 wherein the speech analytics processor stores a segment of the voice signals in a first memory, reduces the voice signals in the segment to a plurality of phonemes, and searches the plurality of phonemes to determine if any of the plurality of phonemes match a second plurality of phonemes associated with the utterance.
7. The method of claim 1 further comprising notifying, upon detecting the utterance, a voice-enabled application of the use of the utterance.
8. An apparatus for monitoring a conversation comprising:
a communications interface adapted to interface with a telecommunications device; and
a control system coupled to the communications interface and adapted to:
couple a voice session comprising voice signals generated during a conversation between a first participant and a second participant to a speech analytics processor to detect in the voice signals a use of an utterance requesting a service from a voice enabled application;
receive from the speech analytics processor an indication that the utterance has been detected in the voice signals and decouple the speech analytics processor from the voice session;
couple the voice session to a speech recognition processor which receives a grammar comprising a plurality of commands for the voice enabled application, wherein the speech recognition processor monitors a media stream subsequent to the use of the utterance for one of the plurality of commands; and
receive from the speech recognition processor an indication that the one of the plurality of commands has been detected and re-couple the speech analytics processor to the voice session.
9. The apparatus of claim 8 wherein the speech analytics processor is further adapted to generate a phonetic index comprising phonemes associated with the voice signals, and determine if one or more phonemes in the phonetic index matches one or more phonemes associated with the utterance.
10. The apparatus of claim 8 wherein to couple the media stream to the speech recognition processor the control system is further adapted to couple the voice session to the speech recognition processor after the utterance has been detected.
11. The apparatus of claim 10 wherein the control system is further adapted to decouple the speech recognition processor from the voice session.
12. The apparatus of claim 10 wherein to couple the voice session to the speech recognition processor the control system is further adapted to couple the voice session to one of a plurality of speech recognition channels associated with a pool of speech recognition channels, and wherein the one of the plurality of speech recognition channels is selected from the pool of speech recognition channels upon detecting the use of the utterance, and the one of the plurality of speech recognition channels is returned to the pool of speech recognition channels upon decoupling the one of the plurality of speech recognition channels from the voice session.
13. The apparatus of claim 8 wherein the speech analytics processor is adapted to store a segment of the voice signals in a first memory, reduce the voice signals in the segment to a plurality of phonemes, and search the plurality of phonemes to determine if any of the plurality of phonemes match a second plurality of phonemes associated with the utterance.
14. The apparatus of claim 8 wherein the control system is further adapted to notify a voice-enabled application of the use of the utterance.
15. A non-transitory computer-usable medium having computer-readable instructions stored thereon for execution by a processor to perform a method comprising:
coupling a voice session comprising voice signals generated during a conversation between a first participant and a second participant to a speech analytics processor to detect in the voice signals a use of an utterance requesting a service from a voice enabled application;
receiving from the speech analytics processor an indication that the utterance has been detected in the voice signals and decoupling the speech analytics processor from the voice session;
coupling the voice session to a speech recognition processor which receives a grammar comprising a plurality of commands for the voice enabled application, wherein the speech recognition processor monitors the voice session subsequent to the use of the utterance for one of the plurality of commands; and
receiving from the speech recognition processor an indication that the one of the plurality of commands has been detected and re-coupling the speech analytics processor to the voice session.
16. The computer-usable medium of claim 15 wherein the method further comprises coupling the speech analytics processor to detect in the voice signals the use of the utterance further comprises generating, by the speech analytics processor, a phonetic index comprising phonemes associated with the voice signals, and determining if one or more phonemes in the phonetic index matches one or more phonemes associated with the utterance.
17. The computer-usable medium of claim 15 wherein coupling the voice session to the speech recognition processor further comprises coupling the voice session to the speech recognition processor after the utterance has been detected.
18. The computer-usable medium of claim 15 wherein the method further comprises decoupling the speech recognition processor from the voice session.
19. The computer-usable medium of claim 15 wherein coupling the voice session to the speech recognition processor comprises coupling the voice session to one of a plurality of speech recognition channels associated with a pool of speech recognition channels, and wherein the one of the plurality of speech recognition channels is selected from the pool of speech recognition channels upon detecting the use of the utterance, and the one of the plurality of speech recognition channels is returned to the pool of speech recognition channels upon decoupling the one of the plurality of speech recognition channels from the voice session.
20. The computer-usable medium of claim 15 wherein the method further comprises notifying, upon detecting the utterance, a voice-enabled application of the use of the utterance.