What are the best text-to-speech ai generation tools that you can run locally?

Mozilla TTS has not been updated in 4 years (at least). The quality of the sound generated is rather poor, or at least I was not able to generate human passable sound using that tool.

tts --text "If you like to use TTS to try a new idea and like to share your experiments with the community, we urge you to use the following guideline for a better collaboration. (If you have an idea for better collaboration, let us know)" \
  --model_name tts_models/en/ljspeech/neural_hmm \
  --vocoder_name vocoder_models/en/ek1/wavegrad \
  --out_path test.wav

OpenVoice, as you very well mentioned, needs to become more mature before it can be used for the purpose I have in mind, as it shares the same poor quality that Mozilla TTS does.

I am currently playing with F5 which even has online tutorials: https://www.youtube.com/watch?v=ASFoTNpkM8o

It seems quite decent. I was able to run it on my local setup as well, which is a big plus. The problem I now face if twofold:

  • I need to find a voice database with male voices in English (have no idea where to find one)
  • I need to then train F5 or whatever tool I use with that voice

As a final step, I will also then need to learn and manipulate said tool to read paragraphs, instead of 1 liners.

2 Likes