├── LICENSE ├── README.md ├── auto_DataLabeling_long.py ├── auto_DataLabeling_re.py └── clean_list.py /LICENSE: -------------------------------------------------------------------------------- 1 | MIT License 2 | 3 | Copyright (c) 2023 AliceNavigator 4 | 5 | Permission is hereby granted, free of charge, to any person obtaining a copy 6 | of this software and associated documentation files (the "Software"), to deal 7 | in the Software without restriction, including without limitation the rights 8 | to use, copy, modify, merge, publish, distribute, sublicense, and/or sell 9 | copies of the Software, and to permit persons to whom the Software is 10 | furnished to do so, subject to the following conditions: 11 | 12 | The above copyright notice and this permission notice shall be included in all 13 | copies or substantial portions of the Software. 14 | 15 | THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR 16 | IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, 17 | FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE 18 | AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER 19 | LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, 20 | OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE 21 | SOFTWARE. 22 | -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- 1 | # auto-VITS-DataLabeling 2 | Simple data labeling script with funasr inside. 使用阿里fanasr进行VITS训练数据标注 3 | -------------------------------------------------------------------------------- /auto_DataLabeling_long.py: -------------------------------------------------------------------------------- 1 | from modelscope.pipelines import pipeline 2 | from modelscope.utils.constant import Tasks 3 | import os 4 | 5 | ''' 6 | inference_pipeline = pipeline( 7 | task=Tasks.auto_speech_recognition, 8 | model='./Model/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch', 9 | ) 10 | 11 | rec_result = inference_pipeline(audio_in='ge_1570_2.wav') 12 | print(rec_result) 13 | # {'text': '欢迎大家来体验达摩院推出的语音识别模型'} 14 | ''' 15 | 16 | parent_dir = "./raw_audio/" 17 | local_dir_model = "./Model/speech_paraformer-large-vad-punc_asr_nat-zh-cn-16k-common-vocab8404-pytorch" 18 | complete_list = [] 19 | filelist = list(os.walk(parent_dir))[0][2] 20 | 21 | if os.path.exists('long_character_anno.txt'): 22 | with open("./long_character_anno.txt", 'r', encoding='utf-8') as f: 23 | for line in f.readlines(): 24 | pt, _, _ = line.strip().split('|') 25 | complete_list.append(pt) 26 | 27 | inference_pipeline = pipeline( 28 | task=Tasks.auto_speech_recognition, 29 | model=local_dir_model 30 | ) 31 | 32 | 33 | for file in filelist: 34 | if file[-3:] != 'wav': 35 | print(f"{file} not supported, ignoring...\n") 36 | continue 37 | print(f"transcribing {parent_dir + file}...\n") 38 | 39 | character_name = file.rstrip(".wav").split("_")[0] 40 | savepth = "./dataset/" + character_name + "/" + file 41 | 42 | if savepth in complete_list: 43 | print(f'{file} is already done, skip!') 44 | continue 45 | 46 | rec_result = inference_pipeline(audio_in=parent_dir + file) 47 | 48 | if 'text' not in rec_result: 49 | print("Text is not recognized，ignoring...\n") 50 | continue 51 | 52 | annos_text = rec_result['text'] 53 | annos_text = '[ZH]' + annos_text.replace("\n", "") + '[ZH]' 54 | annos_text = annos_text + "\n" 55 | line1 = savepth + "|" + character_name + "|" + annos_text 56 | line2 = savepth + "|" + character_name + "|ZH|" + rec_result['text'] + "\n" 57 | with open("./long_character_anno.txt", 'a', encoding='utf-8') as f: 58 | f.write(line1) 59 | with open(f"./barbara.list", 'a', encoding='utf-8') as f: 60 | f.write(line2) 61 | print(rec_result['text']) 62 | print("Done!\n") 63 | 64 | -------------------------------------------------------------------------------- /auto_DataLabeling_re.py: -------------------------------------------------------------------------------- 1 | from modelscope.pipelines import pipeline 2 | from modelscope.utils.constant import Tasks 3 | import os 4 | 5 | ''' 6 | inference_pipeline = pipeline( 7 | task=Tasks.auto_speech_recognition, 8 | model='./Model/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch', 9 | ) 10 | 11 | rec_result = inference_pipeline(audio_in='ge_1570_2.wav') 12 | print(rec_result) 13 | # {'text': '欢迎大家来体验达摩院推出的语音识别模型'} 14 | ''' 15 | 16 | parent_dir = "./raw_audio/" 17 | local_dir_root = "./Model/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch" 18 | target_sr = 44100 19 | 20 | # speaker_annos = [] 21 | # speaker_annos_bert = [] 22 | complete_list = [] 23 | filelist = list(os.walk(parent_dir))[0][2] 24 | 25 | if os.path.exists('long_character_anno.txt'): 26 | with open("./long_character_anno.txt", 'r', encoding='utf-8') as f: 27 | for line in f.readlines(): 28 | pt, _, _ = line.strip().split('|') 29 | complete_list.append(pt) 30 | 31 | inference_pipeline = pipeline( 32 | task=Tasks.auto_speech_recognition, 33 | model=local_dir_root, 34 | ) 35 | 36 | for file in filelist: 37 | if file[-3:] != 'wav': 38 | print(f"{file} not supported, ignoring...\n") 39 | continue 40 | print(f"transcribing {parent_dir + file}...\n") 41 | 42 | character_name = file.rstrip(".wav").split("_")[0] 43 | savepth = "./dataset/" + character_name + "/" + file 44 | 45 | if savepth in complete_list: 46 | print(f'{file} is already done, skip!') 47 | continue 48 | 49 | rec_result = inference_pipeline(audio_in=parent_dir + file) 50 | 51 | if 'text' not in rec_result: 52 | print("Text is not recognized，ignoring...\n") 53 | continue 54 | 55 | annos_text = rec_result['text'] 56 | annos_text = '[ZH]' + annos_text.replace("\n", "") + '[ZH]' 57 | annos_text = annos_text + "\n" 58 | # speaker_annos.append(savepth + "|" + character_name + "|" + annos_text) 59 | line1 = savepth + "|" + character_name + "|" + annos_text 60 | # speaker_annos_bert.append(savepth + "|" + character_name + "|ZH|" + rec_result['text'] + "\n") 61 | line2 = savepth + "|" + character_name + "|ZH|" + rec_result['text'] + "\n" 62 | with open("./long_character_anno.txt", 'a', encoding='utf-8') as f: 63 | f.write(line1) 64 | with open(f"./barbara.list", 'a', encoding='utf-8') as f: 65 | f.write(line2) 66 | print(rec_result) 67 | print("Done!\n") 68 | -------------------------------------------------------------------------------- /clean_list.py: -------------------------------------------------------------------------------- 1 | import os 2 | import re 3 | 4 | new_annos = [] 5 | cleaned_new_annos = [] 6 | 7 | if os.path.exists("./barbara.list"): 8 | with open("./barbara.list", 'r', encoding='utf-8') as f: 9 | long_character_anno = f.readlines() 10 | new_annos += long_character_anno 11 | else: 12 | print('barbara.list cannot be found, please confirm that the path is correct') 13 | exit() 14 | 15 | for line in new_annos: 16 | path, name, lang, text = line.split("|") 17 | text += "\n" if not text.endswith("\n") else "" 18 | if len(text) >= 5: 19 | my_re = re.compile(r'[A-Za-z]', re.S) 20 | res = re.findall(my_re, text) 21 | if len(res): 22 | print(f'Skip non-kanji text : {text}') 23 | else: 24 | cleaned_new_annos.append(path + "|" + name + "|" + lang+ "|" + text) 25 | else: 26 | print(f'skip too short wav : {text}') 27 | 28 | 29 | with open("./clean_barbara.list", 'w', encoding='utf-8') as f: 30 | for line in cleaned_new_annos: 31 | f.write(line) 32 | 33 | print('Done! save as clean_barbara.list') 34 | --------------------------------------------------------------------------------