Unlock instant, AI-driven research and patent intelligence for your innovation.

Method and apparatus for speech recognition

A speech recognition and phoneme recognition technology, applied in speech recognition, speech analysis, instruments, etc., can solve the problems of increasing the difficulty of speech recognition of specific types of named entities, and achieve the effect of improving accuracy

Active Publication Date: 2017-08-04
ALIBABA GRP HLDG LTD
View PDF14 Cites 34 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

The above two situations increase the difficulty of speech recognition for certain types of named entities

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
  • Method and apparatus for speech recognition
  • Method and apparatus for speech recognition
  • Method and apparatus for speech recognition

Examples

Experimental program
Comparison scheme
Effect test

specific Embodiment approach

[0065] In a first implementation manner, the speech recognition based on Pinyin is syllable recognition. The pinyin sequence is a syllable sequence.

[0066] In the first implementation manner, step 120 is specifically performing syllable recognition on the voice of the named entity to be recognized, so as to recognize a syllable sequence that is a syllable recognition result of the voice of the named entity to be recognized.

[0067] That is to say, using the syllable recognition network constituted by the acoustic model and the syllable-based language model to perform syllable recognition on the speech of the named entity to be recognized, so as to recognize a syllable sequence as a syllable recognition result of the speech of the named entity to be recognized. For example, for the speech of the named entity "Zhang San", the syllable sequence "zhang san" is output after the syllable recognition network is used for syllable recognition.

[0068] In the second implementation ...

Embodiment approach

[0105] In the case that the overall edit distance is a weighted average, corresponding to the first or second implementation of step S120, according to the corresponding The weight corresponding to the weight of the syllable sequence edit distance or the weight corresponding to the phoneme sequence edit distance, the edit distance of the Chinese character sequence and the syllable sequence edit distance are weighted, or the edit distance of the Chinese character sequence and the phoneme sequence The edit distance is weighted, and the obtained weighted average is used as the overall edit distance between each candidate named entity in the specific named entity list and the speech of the named entity to be recognized.

[0106] In the case that the overall edit distance is a weighted average, corresponding to the third implementation of step S120, according to the weight, phoneme The weight corresponding to the sequence edit distance and the weight corresponding to the tone seque...

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 application provides a method and apparatus for speech recognition. The method comprises: with Chinese-character-based speech recognition, speech recognition is carried out on to-be-recognized named entity speech to identify a Chinese character sequence of a Chinese character recognition result of the to-be-recognized named entity speech; with Chinese-phonetic-alphabet-based speech recognition, speech recognition is carried out on the to-be-recognized named entity speech to identify a Chinese phonetic alphabet sequence of a Chinese phonetic alphabet identification result of the to-be-recognized named entity speech; according to the identified Chinese character sequence and the identified Chinese phonetic alphabet sequence, similarity degrees between all candidate named entities in a specific named entity list and the to-be-recognized named entity speech are determined; and on the basis of the similarity degrees between all candidate named entities in a specific named entity list and the to-be-recognized named entity speech, a speech recognition result of the to-be-recognized named entity speech is determined from the specific named entity list. Therefore, accuracy of identification of the named entity speech can be improved.

Description

technical field [0001] The present application relates to the field of speech recognition, in particular to a speech recognition method and device. Background technique [0002] Existing speech recognition technologies generally use a speech recognition network composed of a language model and an acoustic model to recognize speech. Wherein, the acoustic model is generated by training the training speech database with a training algorithm, and matching the characteristic parameters of the speech to be recognized with the acoustic model during speech recognition to obtain a recognition result. The language model is generated by analyzing the grammar and semantics of the training text database and training based on statistical models. The language model can combine the knowledge of grammar and semantics to describe the internal relationship between words. [0003] Named Entity (NE) refers to some specific names with entity meanings, such as names, place names, organization nam...

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): G10L15/08G10L15/26G10L15/10
CPCG10L15/08G10L15/10G10L15/26G10L2015/086
Inventor 李宏言
Owner ALIBABA GRP HLDG LTD