Supercharge Your Innovation With Domain-Expert AI Agents!

A speech recognition system and method based on a hybrid acoustic model

An acoustic model and speech recognition technology, which is applied in the field of speech recognition systems based on hybrid acoustic models, can solve problems such as poor robustness of speech recognition, achieve the effect of reducing the error rate of model recognition and improving the performance of speech recognition

Active Publication Date: 2020-11-06
INST OF ACOUSTICS CHINESE ACAD OF SCI +1
View PDF7 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

But time-delay networks and long-short-term memory networks are not as robust as convolutional neural networks in speech recognition

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
  • A speech recognition system and method based on a hybrid acoustic model
  • A speech recognition system and method based on a hybrid acoustic model
  • A speech recognition system and method based on a hybrid acoustic model

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0032] The present invention will be described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0033] A speech recognition system based on a hybrid acoustic model, said system comprising: a signal processing and feature extraction module, a hybrid acoustic model, a pronunciation dictionary, a language model and a decoder;

[0034] Such as figure 2 As shown, the hybrid acoustic model includes: a convolutional neural network and a time-delay and long-short-term memory hybrid neural network, wherein the time-delay and long-short-term memory hybrid neural network is the basic model. The model of the present invention places the convolutional neural network as a feature extraction module before the basic model time delay and long-short-term memory hybrid neural network at the acoustic model level. The input of each time-delay and long-short-term memory hybrid neural network is connected to the front-end convolutional neural network. The ...

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 discloses a voice recognition system and a voice recognition method based on a mixed acoustic model. The voice recognition system comprises a signal processing and character extracting module, a pronouncing dictionary, a voice model, a decoder and the mixed acoustic model, wherein the mixed acoustic model comprises a front-end CNN (Convolutional Neural Network) and a rear-end time delaying and LSTM (Long Short Term Memory) mixed neural network; the CNN which is used as a character extracting module is placed in front of the time delaying and LSTM mixed neural network; a robustness character extracted by the CNN and an original character are spliced to be used as an input character of the rear-end time delaying and LSTM mixed neural network together. The voice recognition system based on the CNN has more robust modeling ability on translation and transformation of characters, the model recognition error rate can be effectively reduced, and voice recognition performance onmultiple task sets can be increased.

Description

technical field [0001] The invention belongs to the field of speech recognition, and in particular relates to a speech recognition system and method based on a hybrid acoustic model. Background technique [0002] Language communication is one of the most natural communication methods for human beings. Human research on computer speech covers speech codec, speech recognition, speech synthesis, speaker recognition, activation words, speech enhancement, etc. Speech recognition is currently the hottest research in these fields. Automatic speech recognition has been put on the agenda long before the invention of computers, and the early vocoder can be regarded as the prototype of speech recognition and synthesis. After decades of research, speech recognition technology has penetrated into all aspects of our lives, and its application range covers smart homes, smart speakers, vehicle interaction, national security and other fields. [0003] At present, the mainstream large-vocab...

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
Patent Type & Authority Patents(China)
IPC IPC(8): G10L15/22G10L15/16
Inventor 徐及程高峰潘接林颜永红
Owner INST OF ACOUSTICS CHINESE ACAD OF SCI
Features
  • R&D
  • Intellectual Property
  • Life Sciences
  • Materials
  • Tech Scout
Why Patsnap Eureka
  • Unparalleled Data Quality
  • Higher Quality Content
  • 60% Fewer Hallucinations
Social media
Patsnap Eureka Blog
Learn More

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

© 2025 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com