Speech recognition method and device

1. A speech recognition method, comprising:

acquiring a text file and extracting a command word from the text file according to a predetermined rule, to obtain a command word list;

generating a corresponding new pronunciation dictionary according to a new command word and performing training to obtain a new language model, wherein the new command word is a command word that is comprised in the command word list but not comprised in a command word library;

merging the new language model into a language model library corresponding to the command word library;

receiving speech and performing speech recognition on the speech according to an acoustic model, a phonation dictionary, and the merged language model library;

obtaining a speech recognition result by means of the speech recognition;

determining whether the speech recognition result is a preset flag command word; and

if the speech recognition result is a preset flag command word, acquiring a text file corresponding to the preset flag command word.

2. The method according to claim 1 , wherein acquiring a text file comprises:

acquiring the text file from a specified address.

3. The method according to claim 1 , wherein extracting a command word from the text file according to a predetermined rule, to obtain a command word list comprises:

reading content of the text file, performing word segmentation on the content, and then selecting the command word from a word segmentation result according to the predetermined rule, to obtain the command word list.

4. A speech recognition method, comprising:

acquiring a text file and extracting a command word from the text file according to a predetermined rule, to obtain a command word list;

comparing the command word list with a command word library, to confirm whether the command word list comprises a new command word, wherein the new command word is a command word that is comprised in the command word list but not comprised in the command word library;

if the command word list comprises the new command word, generating a corresponding new pronunciation dictionary according to the new command word and performing training to obtain a new language model;

merging the new language model into a language model library corresponding to the command word library;

receiving speech and performing speech recognition on the speech according to an acoustic model, a phonation dictionary, and the merged language model library;

determining whether the text file changes; and

if the text file changes, acquiring a changed text file.

5. The method according to claim 4 , wherein acquiring a text file comprises:

acquiring the text file from a specified address.

6. The method according to claim 4 , wherein extracting a command word from the text file according to a predetermined rule, to obtain a command word list comprises:

reading content of the text file, performing word segmentation on the content, and then selecting the command word from a word segmentation result according to the predetermined rule, to obtain the command word list.

7. A speech recognition device, comprising:

a receiver configured to receive speech;

a memory configured to store instruction codes; and

a processor, upon executing the instruction codes, configured to:

acquire a text file;

extract, according to a predetermined rule, a command word from the text file, and obtain a command word list;

compare the command word list with a command word library, and confirm whether the command word list comprises a new command word, wherein the new command word is a command word that is comprised in the command word list but not comprised in the command word library;

if it is determined that the command word list comprises a new command word, generate a corresponding new pronunciation dictionary according to the new command word and perform training to obtain a new language model, and merge the new language model into a language model library corresponding to the command word library;

perform, according to an acoustic model, a phonation dictionary, and the merged language model library, speech recognition on the speech;

after completing the speech recognition, determine whether a speech recognition result is a preset flag command word;

if it is determined that the speech recognition result is a preset flag command word, acquire a text file corresponding to the preset flag command word; and

if it is determined that the speech recognition result is not a preset flag command word, execute an operation.

8. The device according to claim 7 , wherein the processor is configured to:

acquire a text file from a specified address.

9. The device according to claim 7 , wherein the processor is configured to:

read content of the text file;

perform word segmentation on the content; and

select the command word from a word segmentation result according to the predetermined rule, to obtain the command word list.

10. A speech recognition device, comprising:

a receiver configured to receive speech;

a memory configured to store instruction codes; and

a processor, upon executing the instruction codes, configured to:

acquire a text file;

extract, according to a predetermined rule, a command word from the text file, and obtain a command word list;

compare the command word list with a command word library, and confirm whether the command word list comprises a new command word, wherein the new command word is a command word that is comprised in the command word list but not comprised in the command word library;

if it is determined that the command word list comprises a new command word, generate a corresponding new pronunciation dictionary according to the new command word and perform training to obtain a new language model, and merge the new language model into a language model library corresponding to the command word library;

perform, according to an acoustic model, a phonation dictionary, and the merged language model library, speech recognition on the speech;

after completing the speech recognition, determine whether the text file changes;

if it is determined that the text file changes, acquire a changed text file; and

if it is determined that the text file not change, execute an operation.

11. The device according to claim 10 , wherein the processor is further configured to:

acquire a text file from a specified address.

12. The device according to claim 10 , wherein the processor is configured to:

read content of the text file;

perform word segmentation on the content; and

select the command word from a word segmentation result according to the predetermined rule, to obtain the command word list.

13. A speech recognition device, comprising:

a receiver, configured to receive speech;

a memory configured to store instruction codes; and

a processor, upon executing the instruction codes, configured to:

perform, according to an acoustic model, a phonation dictionary, and a language model library that are corresponding to a command word library, speech recognition on the speech received by the speech receiver, to obtain a speech recognition result;

determine whether the speech recognition result obtained by the recognizing unit is a preset flag command word;

it is determined that the speech recognition result is a preset flag command word, acquire a text file corresponding to the preset flag command word;

extract, according to a predetermined rule, a command word from the text file corresponding to the preset flag command word, to obtain a command word list;

compare the command word list with the command word library, and confirm whether the command word list comprises a new command word, wherein the new command word is a command word that is comprised in the command word list but not comprised in the command word library;

if it is determined that the command word list comprises a new command word, generate a corresponding new pronunciation dictionary according to the new command word and perform training to obtain a new language model, and merge the new language model into the language model library corresponding to the command word library.

14. The device according to claim 13 , wherein the processor is configured to:

if it is determined that the speech recognition result is a preset flag command word, acquire a text file from an address corresponding to the preset flag command word, or acquire a text file that is corresponding to the preset flag command word.