wannaphong / ttsmms

TTS with The Massively Multilingual Speech (MMS) project
MIT License
226 stars 37 forks source link
library python text-to-speech

ttsmms

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)

Google colab

Don't forget the TTS model in MMS project use CC-BY-NC license.

Install

pip install ttsmms

Usage

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

  1. download

curl https://dl.fbaipublicfiles.com/mms/tts/lang_code.tar.gz --output lang_code.tar.gz

  1. extract a tar ball archive.

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