Text-to-speech with The Massively Multilingual Speech (MMS) project
This project want to help you for use Text-to-speech model from MMS project in Python.
Support 1,107 Languages! (See support_list.txt)
Don't forget the TTS model in MMS project use CC-BY-NC license.
pip install ttsmms
First, you need to download the model by
from ttsmms import download
dir_path = download("eng","./data") # lang_code, dir for save model
or download file by yourself
Linux/Mac
curl https://dl.fbaipublicfiles.com/mms/tts/lang_code.tar.gz --output lang_code.tar.gz
Linux/Mac
mkdir -p data && tar -xzf lang_code.tar.gz -C data/
and use code in python :D
from ttsmms import TTS
tts=TTS(dir_path) # or "model_dir_path" your path dir that extract a tar ball archive
wav=tts.synthesis("txt")
# output:
# {
# "x":array(wav array),
# "sampling_rate": 16000
# }
tts.synthesis("txt",wav_path="example.wav")
# output: example.wav file