Supercharge Your Innovation With Domain-Expert AI Agents!

Speech recognition decoding efficiency optimization method

An efficiency optimization and speech recognition technology, applied in speech recognition, speech analysis, instruments, etc., can solve problems such as waiting, insufficient memory bandwidth, and affecting the recognition speed of the recognition system, so as to optimize user experience, reduce the number of memory accesses, and improve concurrency The effect of number of paths and recognition speed

Active Publication Date: 2013-04-24
讯飞医疗科技股份有限公司
View PDF4 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Due to insufficient memory bandwidth, it causes waiting time during memory access, which affects the recognition speed of the entire recognition system

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
  • Speech recognition decoding efficiency optimization method
  • Speech recognition decoding efficiency optimization method
  • Speech recognition decoding efficiency optimization method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0020] The present invention adopts an efficiency-optimized frame semi-synchronization method for large-memory speech recognition (especially cloud computing-based speech recognition), so as to save memory access during the recognition process, thereby improving the efficiency of the entire system.

[0021] Compared with the traditional frame synchronization algorithm, the biggest difference of the frame semi-synchronization algorithm is that the Viterbi dynamic programming algorithm is performed every three frames. The implementation process is as follows Figure 4 Shown:

[0022] 1. First plan at time t+1, and update each state as follows:

[0023]

[0024] q t+1 (2)=max[q t (1)+a 12 ,q t (2)+a 22 ]+b 2 (a t+1 )

[0025] q t+1 (3)=max[q t (2)+a 23 ,q t (3)+a 88 ]+b 8 (a t+1 )

[0026] Then plan at time t+2, and the update method of each state is as follows:

[0027]

[0028] q t+2 (2)=max[q t+1 (1)+a 12 ,q t+1 (2)+a 22 ]+b 2 (a t+2 )

[0029] q...

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 relates to a speech reorganization decoding efficiency optimization method. The method includes that for every three frame speech feature vectors, Viterbi dynamic programming is firstly carried out inside an arc. Each arc can output at most three points and corresponding routes; the three points and the routes respectively correspond to output three consecutive and different frames. According to a Viterbi algorithm, the three points and corresponding routes are transmitted to a subsequent node of the arc to carry out a competition. A winner winning on the node can continue to extend to a follow-up arc of the node. For the last frame speech feature vector, the route which is transmitted to the last node of a decoding network and wins on the node is a preferable route. A corresponding word sequence is available by a recollection of the preferable route and the corresponding word sequence is indeed an identification result. The speech reorganization decoding efficiency optimization method saves a visitor volume during the identification process and improves the efficiency of the whole system by adopting a frame synchronization method of efficiency optimization.

Description

technical field [0001] The invention relates to a method for optimizing the decoding efficiency of speech recognition in a continuous speech recognition system, which is used to increase the number of concurrent channels and recognition speed of the speech recognition system based on cloud computing. Background technique [0002] With the popularity of voice input functions and applications on smart terminals such as mobile phones, there are more and more scenarios where users use voice input on smart terminals such as mobile phones. Most of these application scenarios are based on cloud computing. The smart terminal is responsible for recording and audio data compression, and then sends the data to the recognition server in the cloud for recognition, and the recognition result is returned to the smart terminal. For cloud computing-based speech recognition systems, if the number of concurrent channels and recognition speed of a single recognition server can be increased, the...

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 Applications(China)
IPC IPC(8): G10L15/34
Inventor 鹿晓亮赵志伟陈旭尚丽吴晓如于振华潘青华
Owner 讯飞医疗科技股份有限公司
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