Self-adaptive training method and system for acoustic model

A training method and technology for acoustic models, applied in the field of adaptive training methods and systems for acoustic models, can solve problems such as damage to seed model information, reduce overfitting, reduce overfitting, and improve recognition rates. Effect

Active Publication Date: 2018-11-02
SOUNDAI TECH CO LTD
View PDF4 Cites 10 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0003] The existing acoustic model adaptive technology directly uses the adaptive data to adjust the weight of the neural network, but the algorithm has the problem of model overfittin

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Self-adaptive training method and system for acoustic model
  • Self-adaptive training method and system for acoustic model
  • Self-adaptive training method and system for acoustic model

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0040] This disclosure provides an adaptive training method and system for an acoustic model, combining the advantages of LHT and KL divergence, using adaptive data to retrain the neural network model, to achieve the purpose of slowing down the mismatch between training data and scene data, and ensuring Improve the accuracy of the model during the adaptation process.

[0041] Before describing a solution to a problem, it is helpful to define some specific vocabulary definitions.

[0042] KLD Kullback-Leibler divergence KL divergence

[0043] LHT Linear Hidden Transformations linear hidden transformation

[0044] HMM Hidden Markov Model Hidden Markov Model

[0045] DNN Deep Neural Networks Deep Neural Networks

[0046] In order to make the purpose, technical solutions and advantages of the present disclosure clearer, the present disclosure will be further described in detail below in conjunction with specific embodiments and with reference to the accompanying drawings.

[0...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention provides a self-adaptive training method for an acoustic model. The self-adaptive training method comprises the steps of: step S1, extracting speech features, and taking the speech feature as input for training and generating a seed model, so as to obtain an objective function; step S2, adjusting a network structure of the seed model, and adding a linear layer; step S3, adding a KL divergence regular term on the basis of the objective function; step S4, training the linear layer, and estimating weight and offset of a hidden linear layer by reusing a back propagation algorithm; step S5, and completing the training, and outputting a self-adaptive model. Since LHT can map scene data, and KL divergence can alleviate the over-fitting phenomenon, the self-adaptive training method can ensure that the occurrence of the over-fitting phenomenon in the neural network training process can be reduced in the case of few self-adaptive data, and the identification rate of the scene datais improved.

Description

technical field [0001] The present disclosure relates to the field of speech recognition, in particular to an adaptive training method and system for an acoustic model. Background technique [0002] Automatic speech recognition is an important direction of artificial intelligence application, and has developed into a new high-tech industry with broad prospects. In recent years, with the rise of industries such as smart home and car navigation, far-field speech recognition technology has received keen attention. Far-field speech recognition systems usually include front-end signal processing and back-end speech recognition modules. The front-end part aims to process speech containing noise and reverberation into "Clean" voice. The back-end part is the same as the general speech recognition system, the purpose is to recognize the processed "clean" speech as text. In order to get a better recognition effect, the back-end speech recognition needs to match the front-end noise ...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
IPC IPC(8): G10L15/06G10L15/16
CPCG10L15/063G10L15/16G10L15/26
Inventor 谭应伟陈孝良冯大航苏少炜常乐
Owner SOUNDAI TECH CO LTD
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap