Language model customization in speech recognition for speech analytics

Patent metadata

Field	Value
Publication number	US-10186255-B2
Application number	US-201615247645-A
Country	US
Kind code	B2
Filing date	Aug 25, 2016
Priority date	Jan 16, 2016
Publication date	Jan 22, 2019
Grant date	Jan 22, 2019

Abstract

A method for generating a language model for an organization includes: receiving, by a processor, organization-specific training data; receiving, by the processor, generic training data; computing, by the processor, a plurality of similarities between the generic training data and the organization-specific training data; assigning, by the processor, a plurality of weights to the generic training data in accordance with the computed similarities; combining, by the processor, the generic training data with the organization-specific training data in accordance with the weights to generate customized training data; training, by the processor, a customized language model using the customized training data; and outputting, by the processor, the customized language model, the customized language model being configured to compute the likelihood of phrases in a medium.

First claim

What is claimed is:

1. A method for performing speech recognition of interactions with an organization, comprising:
training a customized language model for the organization by:
receiving, by a processor, organization-specific training data comprising a plurality of organization-specific phrases;
receiving, by the processor, generic training data comprising a plurality of generic phrases;
computing, by the processor, a plurality of similarities between the generic training data and the organization-specific training data;
assigning, by the processor, a plurality of weights to the generic training data in accordance with the computed similarities;
combining, by the processor, the generic training data with the organization-specific training data in accordance with the weights to generate customized training data;
training, by the processor, the customized language model using the customized training data; and
outputting, by the processor, the customized language model, the customized language model being configured to compute the likelihood that an input phrase will appear in a communication medium in an interaction with the organization; and

receiving input speech from an interaction between a customer and a contact center of the organization;
transcribing the received input speech, by an automatic speech recognition engine configured with the customized language model, to generate a transcript of the input speech; and
performing voice analytics on the transcript of the input speech.

2. The method of claim 1 , wherein the organization-specific training data comprise in-medium data and out-of-medium data.

3. The method of claim 2 , wherein the in-medium data are speech recognition transcript text and the out-of-medium data are non-speech text.

4. The method of claim 1 , wherein the organization-specific training data does not include in-medium data.

5. The method of claim 1 , wherein the assigning the plurality of weights to the generic training data comprises:
partitioning the generic training data into a plurality of partitions in accordance with the computed similarities;
associating a partition similarity with each of the partitions, the partition similarity corresponding to the average similarity of the data in the partition; and
assigning a desired weight to each partition, the desired weight corresponding to the partition similarity of the partition.

6. The method of claim 5 , wherein the desired weight of a partition is exponentially decreasing with decreasing partition similarity.

7. The method of claim 1 , further comprising:
receiving organization-specific in-medium data;
combining the organization-specific in-medium data with the generic training data and the organization-specific training data to generate the customized training data; and
retraining the language model in accordance with the customized training data.

8. A system configured to perform speech recognition of interactions with an organization, the system comprising:
a processor;
memory coupled to the processor and storing instructions that, when executed by the processor, cause the processor to:
receive organization-specific training data comprising a plurality of organization-specific phrases;
receive generic training data comprising a plurality of generic phrases;
compute a plurality of similarities between the generic training data and the organization-specific training data;
assign a plurality of weights to the generic training data in accordance with the computed similarities;
combine the generic training data with the organization-specific training data in accordance with the weights to generate customized training data;
train a customized language model using the customized training data; and
output the customized language model, the customized language model being configured to compute the likelihood that an input phrase will appear in a communication medium in an interaction with the organization; and

a speech recognition module configured to:
receive input speech from an interaction between a customer and a contact center of the organization;
transcribe the received input speech, by an automatic speech recognition engine configured with the customized language model, to generate a transcript of the input speech; and
perform voice analytics on the transcript of the input speech.

9. The system of claim 8 , wherein the organization-specific training data comprise in-medium data and out-of-medium data.

10. The system of claim 9 , wherein the in-medium data are speech recognition transcript text and the out-of-medium data are non-speech text.

11. The system of claim 8 , wherein the organization-specific training data does not include in-medium data.

12. The system of claim 8 , wherein the memory further stores instructions that, when executed by the processor, cause the processor to assign the plurality of weights to the generic training data by:
partitioning the generic training data into a plurality of partitions in accordance with the computed similarities;
associating a partition similarity with each of the partitions, the partition similarity corresponding to the average similarity of the data in the partition; and
assigning a desired weight to each partition, the desired weight corresponding to the partition similarity of the partition.

13. The system of claim 12 , wherein the desired weight of a partition is exponentially decreasing with decreasing partition similarity.

14. The system of claim 8 , wherein the memory further stores instructions that, when executed by the processor, cause the processor to:
receive organization-specific in-medium data;
combine the organization-specific in-medium data with the generic training data and the organization-specific training data to generate the customized training data; and
retrain the language model in accordance with the customized training data.

Assignees

• Genesys Telecommunications Laboratories Inc

Inventors

• Tapuhi Tamir
• Lev-Tov Amir
• Faizakof Avraham
• Konig Yochai

Classifications

• G06N3/006

  based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO] · CPC title

• G10L15/063 Primary

  Training · CPC title

• G10L15/183

  using context dependencies, e.g. language models · CPC title

• G10L2015/088

  Word spotting · CPC title

• G06N20/00 Primary

  Machine learning · CPC title