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

Speech recognition model training method, system, mobile terminal and storage medium

A speech recognition model and training method technology, applied in speech recognition, speech analysis, instruments, etc., can solve the problems of long time and low training efficiency, and achieve the effect of improving efficiency, reducing model training time, and reducing labor costs.

Active Publication Date: 2020-05-12
XIAMEN KUAISHANGTONG TECH CORP LTD
View PDF5 Cites 7 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] The purpose of the embodiments of the present invention is to provide a speech recognition model training method, system, mobile terminal and storage medium, aiming to solve the problems of low training efficiency and time-consuming in the existing speech recognition model training method

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 model training method, system, mobile terminal and storage medium
  • Speech recognition model training method, system, mobile terminal and storage medium
  • Speech recognition model training method, system, mobile terminal and storage medium

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0050] see figure 1 , is a flow chart of the speech recognition model training method provided in the first embodiment of the present invention, including steps:

[0051] Step S10, acquiring sample speech, sample text corresponding to the sample speech, and text corpus, and constructing a text dictionary based on the sample text and the text corpus;

[0052] Wherein, the sample speech is a language to be recognized by the speech recognition model, such as Cantonese or Hokkien, and the sample text is expressed in Mandarin, and there is a one-to-one correspondence between the sample speech and the sample text;

[0053] Specifically, through the acquisition of the sample speech and sample text, a corresponding data set is constructed, and 20% of the data in the data set are randomly selected as the test set;

[0054] In this step, before the step of constructing a text dictionary according to the sample text and the text corpus, the method includes:

[0055] Deleting the special ...

Embodiment 2

[0070] see figure 2 , is a flow chart of the speech recognition model training method provided by the second embodiment of the present invention, including steps:

[0071] Step S11, acquiring sample speech, sample text and text corpus corresponding to the sample speech;

[0072] Step S21, traversing the local pre-stored training text, adding all non-repetitive characters to the text dictionary to build a character set;

[0073] Among them, each character is represented by a corresponding unique ID;

[0074] Step S31, the characters in the sample text and the text corpus are replaced with corresponding IDs according to the character set, and the characters in the text corpus that are not in the character set are represented by a first identifier;

[0075] Among them, the first identification can adopt expressed in a manner;

[0076] Step S41, adding the first identification to the character set, and using the current maximum ID of the character set plus 1 to represent; ...

Embodiment 3

[0113] see Figure 5 , is a schematic structural diagram of the speech recognition model training system 100 provided by the third embodiment of the present invention, including: a dictionary construction module 10, a vector calculation module 11, a model training module 12 and a model integration module 13, wherein:

[0114] Dictionary construction module 10 is used to obtain sample speech, sample text and text corpus corresponding to the sample speech, and construct a text dictionary according to the sample text and the text corpus.

[0115] Wherein, the dictionary construction module 10 is also used for:

[0116] Traversing the local pre-stored training text, adding all non-repetitive characters to the text dictionary to build a character set, and each character is represented by a corresponding unique ID;

[0117] replacing the characters in the sample text and the text corpus with corresponding IDs according to the character set;

[0118] Representing charac...

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 speech recognition model training method, a system, a mobile terminal and a storage medium. The speech recognition model training method comprises the steps: obtaining samplespeech, a sample text corresponding to the sample speech, and a text corpus, and constructing a text dictionary according to the sample text and the text corpus; performing feature extraction on the sample speech to obtain speech features, and performing vector calculation on the speech features to obtain a probability vector; performing loss calculation according to the probability vector and thetext dictionary to obtain model total loss, and propagating the model total loss in a speech model until a speech recognition model converges; and training a language model according to the text corpus, and integrating the trained language model into the speech recognition model. According to the method, a pronunciation dictionary does not need to be constructed, the labor cost is reduced, the model training time is shortened, all parameters are updated at the same time by adopting an end-to-end architecture training mode, and the model training efficiency and the subsequent speech recognition efficiency are improved.

Description

technical field [0001] The invention belongs to the technical field of speech recognition, and in particular relates to a speech recognition model training method, system, mobile terminal and storage medium. Background technique [0002] Speech recognition research has a history of several decades. Speech recognition technology mainly includes four parts: acoustic model modeling, language model modeling, pronunciation dictionary construction, and decoding. Each part can become a separate research direction, and compared with image And text, the difficulty of collecting and labeling speech data is also greatly increased, so building a complete speech recognition model training system is a very time-consuming and extremely difficult task, which greatly hinders the development of speech recognition technology. With the research and development of artificial intelligence technology, especially deep learning, some end-to-end speech recognition algorithms have been proposed. Compa...

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): G10L15/06G10L15/16G10L15/02
CPCG10L15/063G10L15/16G10L15/02
Inventor 徐敏肖龙源李稀敏蔡振华刘晓葳谭玉坤
Owner XIAMEN KUAISHANGTONG TECH CORP LTD