提取语音输出文本唯一出现ID

WiTiAi

【1】采用halcon软件提取语音输出文本唯一出现ID

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. * 在pathtranscript里面查找对应的wav对应的输出,找出唯一出现过的字母或者文字
    
19. WavNameOutput := ''
    
20. for k:=0 to |pathtranscriptWords|-1 by 1
    
21.     scriptWord := pathtranscriptWords[k]
    
22.     tuple_split(scriptWord, ' ', Substrings2)
    
23.     tuple_length(Substrings2, Length1)
    
24.     if(Length1>0)
    
25.         tuple_split(scriptWord, Substrings2[0], Substrings3)
    
26.         tuple_strlen(Substrings3, Length)
    
27.         tuple_substr(Substrings3, 1, Length-1, Substring)
    
28.         tuple_split(Substring[0], '\n', Substrings)
    
29.         WavNameOutput := WavNameOutput + Substrings[0]
    
30.     endif
    
31. endfor
    
32. * WriteTxt (FilePath + WavName + '.txt', WavNameOutput)
    
33. * WavInputOut := WavInputOut + WavFile + ',' + FilePath + WavName + '.txt' + '\n'
    
34. *
    
35. * resPath := 'D:/2-LearningCode/902-ASR/AISHELL-1/data_aishell/'
    
36. * WriteTxt (resPath + 'ASRInfo.txt', WavInputOut)
    
37. UniqueWord := ''
    
38. tuple_strlen(WavNameOutput, Length2)
    
39. for i:= 0 to Length2-1 by 1
    
40.     tuple_substr(WavNameOutput, i, i, CurChar)
    
41.     tuple_strchr(UniqueWord, CurChar, Position)
    
42.     if(Position==-1 and CurChar != ' ')
    
43.         UniqueWord := UniqueWord + CurChar
    
44.     endif
    
45. endfor
    
46. * WriteTxt (FilePath + WavName + '.txt', WavNameOutput)
    
47. File_Name := 'D:/2-LearningCode/902-ASR/AISHELL-1/data_aishell/LabelWav.json'
    
48. file_exists(File_Name,FileExits)
    
49. if(FileExits)
    
50.     delete_file(File_Name)
    
51.     open_file(File_Name,'append',FileHandle)
    
52. else
    
53.     open_file(File_Name,'output',FileHandle)
    
54. endif
    
55. fwrite_string(FileHandle, '[' + '\n')
    
56. tuple_strlen(UniqueWord, Length3)
    
57. for i:= 0 to Length3-1 by 1
    
58.     tuple_substr(UniqueWord, i, i, Substring1)
    
59.     fwrite_string(FileHandle, '"' + Substring1 + '"' + '\n')
    
60. endfor
    
61. fwrite_string(FileHandle,  '"' + ' ' + '"'+ '\n')
    
62. fwrite_string(FileHandle, ']' + '\n')
    
63. close_file (FileHandle)

复制代码

输出如下：

1. [
   
2.   "_",
   
3.   "'",
   
4.   "a",
   
5.   "b",
   
6.   "c",
   
7.   "d",
   
8.   "e",
   
9.   "f",
   
10.   "g",
    
11.   "h",
    
12.   "i",
    
13.   "j",
    
14.   "k",
    
15.   "l",
    
16.   "m",
    
17.   "n",
    
18.   "o",
    
19.   "p",
    
20.   "q",
    
21.   "r",
    
22.   "s",
    
23.   "t",
    
24.   "u",
    
25.   "v",
    
26.   "w",
    
27.   "x",
    
28.   "y",
    
29.   "z",
    
30.   " "
    
31. ]

复制代码