Apparatus and Methods Using a Pattern Matching Speech Recognition Engine to Train a Natural Language Speech Recognition Engine

• US 20130332158A1
• Filed: 06/08/2012
• Published: 12/12/2013
• Est. Priority Date: 06/08/2012
• Status: Active Grant

• Current Assignee: nVoq, Inc.
• Original Assignee: nVoq, Inc.
• Inventors: Corfield, Charles, Marquette, Brian
• Granted Patent: US 9,767,793 B2
• Time in Patent Office: Days
• US Class Current: 704/235
• CPC Class Codes: G10L 15/07 to the speaker
• CPC Class Codes: G10L 15/32 Multiple recognisers used in sequence or in parallel; Score combination systems therefor, e.g. voting systems

Abstract

The technology of the present application provides a speech recognition system with at least two different speech recognition engines or a single engine speech recognition engine with at least two different modes of operation. The first speech recognition being used to match audio to text, which text may be words or phrases. The matched audio and text is used by a training module to train a user profile for a natural language speech recognition engine, which is at least one of the two different speech recognition engines or modes. An evaluation module evaluates when the user profile is sufficiently trained to convert the speech recognition engine from the first speech recognition engine or mode to the natural language speech recognition or mode.

Claims

1. A method performed on at least one processor for training a user profile for a natural language speech recognition engine, the method comprising the steps of:

providing an audio file wherein the audio file contains a pronunciation of a client;

recognizing the audio file as text data using a first speech recognition engine, wherein the first speech recognition engine is not a natural language speech recognition engine;

linking the audio file and the text data;

initially training a user profile of the natural language speech recognition engine using the linked audio file and text data, wherein the user profile of the natural language speech recognition engine is initially trained using audio recognized by the first speech recognition engine.

2. The method of claim 1 wherein the first speech recognition engine is a pattern matching speech recognition engine.

3. The method of claim 1 further comprising the step of converting the recognition engine from the first speech recognition engine to the natural language speech recognition engine.

4. The method of claim 3 wherein the converting step comprises the step of evaluating whether the user profile is sufficiently trained.

5. The method of claim 4 wherein the evaluating step determines the user profile is sufficiently trained when a predetermined percentage of words in the first recognition engine vocabulary are trained.

6. The method of claim 4 wherein the evaluating step determines the user profile is sufficiently trained when a plurality of phonemes have been spoken a number of times.

7. The method of claim 5 wherein the evaluating step further comprises determining whether a plurality of phonemes have been spoken a number of times.

8. The method of claim 4 wherein the evaluating step determines the user profile is sufficiently trained when a client has used the first recognition engine a predetermined number of hours.

9. The method of claim 3 wherein the conversion step comprises:

ceasing operation of the first recognition engine; and

commencing operation of the natural language speech recognition engine.

10. The method of claim 3 wherein the conversion step comprises:

phasing out the operation of the first recognition engine; and

phasing in the operation of the natural language speech recognition engine.

11. An apparatus for speech recognition comprising:

a first speech recognition engine wherein the first speech recognition engine is not a natural language speech recognition engine, the first speech recognition engine to receive audio from a client and match the received audio to text;

a memory coupled to the first speech recognition engine to store the received audio and the text, the memory configured to store a user profile;

a natural language speech recognition engine, the natural language speech recognition engine coupled to the memory and capable of fetching at least the user profile;

a training module, the training module coupled to the memory and the natural language speech recognition engine,

wherein the training module uses the received audio and the text at least to initially train the user profile.

12. The apparatus of claim 11 wherein the first speech recognition engine is a pattern matching recognition engine.

13. The apparatus of claim 12 further comprising an evaluation module, the evaluation module configured to determine whether the user profile is sufficiently trained to enable the natural language speech recognition engine.

14. The apparatus of claim 13 wherein the evaluation module determines the user profile is sufficiently trained when a predetermined percentage of the pattern matching speech recognition engine vocabulary has been trained.

15. The apparatus of claim 13 wherein the evaluation module determines the user profile is sufficiently trained when a plurality of phonemes have been spoken a number of times.

16. The apparatus of claim 14 wherein the evaluation module determines the user profile is sufficiently trained when a plurality of phonemes have been spoken a number of times.

17. The apparatus of claim 13 wherein the evaluation module determines the user profile is sufficiently trained when it determines the first speech recognition engine has been used a predetermined amount of time.