Speech recognition over packet networks

• US 6,195,636 B1
• Filed: 02/19/1999
• Issued: 02/27/2001
• Est. Priority Date: 02/19/1999
• Status: Expired due to Term

• Current Assignee: Telogy Networks, Inc. (Texas Instruments, Inc.)
• Original Assignee: Texas Instruments, Inc.
• Inventors: Mladenovic, Zoran, Morgan, Edward B., Kosanovic, Bogdan R., Kumar, Negendra, Crupi, Joseph A.
• Primary Examiner(s): Hudspeth, David
• Assistant Examiner(s): Chawan, Vijay B
• Application Number:
  US09/253,017
• Time in Patent Office: 739 Days
• Field of Search: 704/233, 704/241, 704/243, 704/231, 704/256, 704/270, 704/275, 704/201
• US Class Current: 704/231
• CPC Class Codes: G10L 15/30 Distributed recognition, e.g. in client-server systems, for mobile phones or network applications

Abstract

In a system in which user equipment is connected to a packet network and a speech recognition application server is also connected to the packet network for performing speech recognition on speech data from the user equipment, a speech recognition system selectively performs feature extraction at a user end before transmitting speech data to be recognized. The feature extraction is performed only for speech which is to be recognized.

Claims

1. A speech recognition system comprising:

user equipment connected to a first side of a packet network for accepting speech input; and

for performing partial speech recognition of said speech input, and;

a speech recognition application server connected to a second side of said packet network for performing speech recognition on speech data corresponding to said speech input wherein;

said speech data includes feature extraction data from said partial recognition which is provided across said packet network to said speech recognition application server according to a first protocol and includes decompressed speech transmitted to the speech recognition application server via the packet network according to a second protocol,

said speech recognition application server performing said speech recognition based upon said feature extraction data and said decompressed speech.

2. A system as in claim 1 wherein the partial speech recognition performed by the user equipment includes feature extraction of features included within the speech.

3. A system as in claim 2 wherein the feature extraction includes at least one of:

cepstral analysis, spectral analysis and perceptual analysis.

4. A system as in claim 1 wherein the extracted features are compressed prior to being transmitted over the packet network.

5. A system as in claim 4 wherein the features are compressed using at least one of:

linear quantization and vector quantization.

6. A speech recognition system according to claim 1, further comprising:

means for generating a recognition signal if speech recognition is to be performed by said speech recognition application server; and

means for receiving said recognition signal and for inhibiting said feature extraction and said transmission of extracted features unless said recognition signal is received.

7. In a system in which user equipment is connected to a first side of a packet network and a speech recognition application server is connected to a second side of said packet network for performing speech recognition on speech data, a method for implementing speech recognition across a packet network, comprising:

inputting speech to said user equipment;

extracting features from said speech by selectively performing partial speech recognition on said speech at said user equipment side of said packet network;

transmitting said extracted features across said network to said speech recognition application server according to a first codec selected for transmission of extracted features, and

transmitting said speech across said network to said speech recognition application server according to a second codec selected for transmission of speech, and

providing said speech and said extracted features as speech data to said speech recognition application server for recognition.

8. A method as in claim 7 wherein the partial speech recognition performed by the user equipment includes feature extraction of features included within the speech.

9. A method as in claim 8 wherein the feature extraction includes at least one of:

cepstral analysis, spectral analysis and partial analysis.

10. A method as in claim 7 further comprising:

compressing the extracted features prior to transmitting them over the packet network.

11. A method as in claim 10 further comprising:

compressing the features using at least one of linear quantization and vector quantization.

12. A method for implementation of speech recognition across a packet network according to claim 7, further comprising the steps of:

providing an indication that speech recognition is to be performed at said second side of said packet network by said speech recognition application server;

receiving said indication at said first side of said packet network;

wherein

said feature extraction is performed only upon receipt of said indication.