GitHub - Cosmos-Break/asr: 沪语（上海话）ASR（语音识别）模型

使用的工具库

Huggingface的transformers和huggingsound。

包含一个沪语ASR模型(沪语语音->沪语转写文本)和机器翻译模型(沪语转写文本->普通话文本)。

包含Magichub开源数据集、喜马拉雅，shh.dict.cn，中国语言网爬取的数据集、讯飞TTS生成的wav数据。

train.py 用于训练ASR模型

train_translation.py 用于训练MT模型

使用fastapi进行整个模型的部署，运行run_service.sh部署。

Name		Name	Last commit message	Last commit date
Latest commit History 61 Commits
.vscode		.vscode
Shanghai_Dialect_Conversational_Speech_Corpus		Shanghai_Dialect_Conversational_Speech_Corpus
Shanghai_Dialect_Dict		Shanghai_Dialect_Dict
Shanghai_Dialect_Scripted_Speech_Corpus_Daily_Use_Sentence		Shanghai_Dialect_Scripted_Speech_Corpus_Daily_Use_Sentence
Shanghai_Dialect_Ximalaya		Shanghai_Dialect_Ximalaya
Shanghai_Dialect_Zhongguoyuyan		Shanghai_Dialect_Zhongguoyuyan
app		app
.gitignore		.gitignore
README.md		README.md
change_vocab_size.py		change_vocab_size.py
collect_vocab.py		collect_vocab.py
make_requests.py		make_requests.py
push.sh		push.sh
requirements.txt		requirements.txt
run.sh		run.sh
run_service.sh		run_service.sh
split_audio_1.py		split_audio_1.py
split_audio_2.py		split_audio_2.py
split_audio_3.py		split_audio_3.py
split_audio_4.py		split_audio_4.py
train.py		train.py
train_translation.py		train_translation.py
transcribe.py		transcribe.py
translation.py		translation.py