提取AISHELL-1输入输出

WiTiAi

提取AISHELL-1输入输出
【1】采用halcon进行提取

1. dev_update_off()
   
2. dev_set_draw('margin')
   
3. pathwav:='D:/2-LearningCode/902-ASR/AISHELL-1/data_aishell/wav/train'
   
4. pathtrans := 'D:/2-LearningCode/902-ASR/AISHELL-1/data_aishell/transcript'
   
5. pathtranscript:='D:/2-LearningCode/902-ASR/AISHELL-1/data_aishell/transcript/aishell_transcript_v0.8.txt'
   
6. pathtranscriptWords := []
   
7. open_file (pathtranscript, 'input', FileHandle)
   
8. IsEOF := 0
   
9. nums := []
   
10. k:=0
    
11. while (IsEOF == 0)
    
12.   fread_line (FileHandle, Number, IsEOF)
    
13.   pathtranscriptWords[k] := Number
    
14.   k:=k+1
    
15. endwhile
    
16. close_file (FileHandle)
    
17. *      
    
18. WavInputOut := ''
    
19. list_files(pathwav, 'directories', Files)
    
20. for i:= 0 to |Files|-1 by 1
    
21.     FilePath:=Files[i]
    
22.     list_image_files(FilePath, 'wav', [], WavFiles)
    
23.     for j:= 0 to |WavFiles|-1 by 1
    
24.         WavFile := WavFiles[j]
    
25.         tuple_split(WavFile, '/', Substrings)
    
26.         WavNameExt := Substrings[|Substrings|-1]
    
27.         tuple_split(WavNameExt, '.wav', Substrings1)
    
28.         WavName:= Substrings1[0]
    
29.         * 在pathtranscript里面查找对应的wav对应的输出
    
30.         WavNameOutput := ''
    
31.         for k:=0 to |pathtranscriptWords|-1 by 1
    
32.             scriptWord := pathtranscriptWords[k]
    
33.             tuple_split(scriptWord, ' ', Substrings2)
    
34.             tuple_length(Substrings2, Length1)
    
35.             if(Length1>0)
    
36.                 if(Substrings2[0]==WavName)
    
37.                     tuple_split(scriptWord, WavName, Substrings3)
    
38.                     tuple_strlen(Substrings3, Length)
    
39.                     tuple_substr(Substrings3, 1, Length-1, Substring)
    
40.                     WavNameOutput := Substring
    
41.                     break
    
42.                 endif
    
43.             endif
    
44.         endfor
    
45.         WriteTxt (FilePath + WavName + '.txt', WavNameOutput)
    
46.         WavInputOut := WavInputOut + WavFile + ',' + FilePath + WavName + '.txt' + '\n'
    
47.     endfor
    
48. endfor
    
49. *
    
50. resPath := 'D:/2-LearningCode/902-ASR/AISHELL-1/data_aishell/'
    
51. WriteTxt (resPath + 'ASRInfo.txt', WavInputOut)
    

复制代码

写txt

1. file_exists(File_Name,FileExits)
   
2. if(FileExits)
   
3.     delete_file(File_Name)
   
4.     open_file(File_Name,'append',FileHandle)
   
5. else
   
6.     open_file(File_Name,'output',FileHandle)
   
7. endif
   
8. fwrite_string(FileHandle, outputstr + '\n')
   
9. close_file (FileHandle)
   
10. return ()

复制代码

得到结果如下：

1. D:/2-LearningCode/902-ASR/AISHELL-1/data_aishell/wav/train/S0002/BAC009S0002W0122.wav,D:\2-LearningCode\902-ASR\AISHELL-1\data_aishell\wav\train\S0002\BAC009S0002W0122.txt
   
2. D:/2-LearningCode/902-ASR/AISHELL-1/data_aishell/wav/train/S0002/BAC009S0002W0123.wav,D:\2-LearningCode\902-ASR\AISHELL-1\data_aishell\wav\train\S0002\BAC009S0002W0123.txt
   
3. D:/2-LearningCode/902-ASR/AISHELL-1/data_aishell/wav/train/S0002/BAC009S0002W0124.wav,D:\2-LearningCode\902-ASR\AISHELL-1\data_aishell\wav\train\S0002\BAC009S0002W0124.txt
   
4. D:/2-LearningCode/902-ASR/AISHELL-1/data_aishell/wav/train/S0002/BAC009S0002W0125.wav,D:\2-LearningCode\902-ASR\AISHELL-1\data_aishell\wav\train\S0002\BAC009S0002W0125.txt
   
5. D:/2-LearningCode/902-ASR/AISHELL-1/data_aishell/wav/train/S0002/BAC009S0002W0126.wav,D:\2-LearningCode\902-ASR\AISHELL-1\data_aishell\wav\train\S0002\BAC009S0002W0126.txt

复制代码