MiraTTS

MiraTTS

This is the model for the MiraTTS repository. MiraTTS is a high quality TTS model that can generate clear and realistic speech at speeds as fast as 100x realtime.

Key benefits

• Incredibly fast: Over 100x realtime by using Lmdeploy and batching.
• High quality: Generates clear and crisp 48khz audio outputs which is much higher quality then most models.
• Memory efficient: Works within 6gb vram.
• Low latency: Latency can be low as 100ms.
• Voice cloning: Can voice clone any voice with good quality.

Random samples, non cherry picked: <audio controls> <source src="https://huggingface.co/YatharthS/MiraTTS/resolve/main/example2.wav" type="audio/wav"> Your browser does not support the audio element. </audio> <audio controls> <source src="https://huggingface.co/YatharthS/MiraTTS/resolve/main/example3.wav" type="audio/wav"> Your browser does not support the audio element. </audio> <audio controls> <source src="https://huggingface.co/YatharthS/MiraTTS/resolve/main/example1.wav" type="audio/wav"> Your browser does not support the audio element. </audio>

Thanks to Gapeleon for creating a great space for this model, you can try it here: https://huggingface.co/spaces/Gapeleon/Mira-TTS

If you find this model/code helpful, please give a like or star. Thank you.

Please check out the github repo for usage and finetuning notebooks.