The aforementioned Bark by SunoAI always sounds like a foreigner speaking Chinese, which is not authentic and pure. Let's play with Baidu Paddle--PaddleSpeech, which is an easy-to-use all-in-one speech toolbox that supports speech recognition, speech synthesis, voiceprint recognition, sound classification, speech translation, punctuation recovery, and voice wake-up. Development work in multiple directions.
Project address: https://github.com/PaddlePaddle/PaddleSpeech
It is strongly recommended to use python>=3.8 and paddlepaddle>=2.5.0 to install PaddleSpeech in Linux.
1. Install
the # cpu version of the example
conda activate sadtalker
pip install paddlepaddle -i https://mirror.baidu.com/pypi/simple
# pip install paddlespeech
git clone https://github.com/PaddlePaddle/PaddleSpeech.git
cd PaddleSpeech
pip install pytest-runner
pip install .
2. Automatic speech recognition
mkdir /home/project/PaddleSpeech
wget -c https://paddlespeech.bj.bcebos.com/PaddleAudio/zh.wav
paddlespeech asr --lang zh --input zh.wav
# Use python code
>> > from paddlespeech.cli.asr.infer import ASRExecutor
>>> asr = ASREx