I want to create TTS app.
I searched and find many libs for this but they are not use my language (Uzbek). If I usethem, result is not correct.
So i want to make own lib for TTS with my own voice. But i can not find way to make this. Who can help. How i make this?
Advertisement
Answer
For Uzbek language you can use eSpeak-NG with py-espeak-ng Python module: https://pypi.org/project/py-espeak-ng/
According to eSpeak documentation, it supports 158 languages, including Uzbek.
Here is example:
from espeakng import ESpeakNG
esng = ESpeakNG()
esng.voice = 'uz'
esng.say('Salom Dunyo')
But if you want to use your own voice, take a look at Real-time Voice Cloning.
You can take pretrained model and retrain it on your own samples.
Also here is a good manual about setting up TTS with Real-Time Voice Cloning environment in Google Colab:
https://medium.com/wavy-engineering/text-to-speech-with-real-time-voice-cloning-16346127742