DEV Community

Jeremy Xiao
Jeremy Xiao

Posted on

Using a TTS model directory: what I check before testing a voice

#ai

Disclosure: I work on a small project called BestTTS, so this is a first-person note rather than a neutral review.

BestTTS is a free directory and comparison site for text-to-speech and speech AI models. It is a directory, not a TTS generator: you cannot type text into it and get audio back.

What I check before I run any tests

  1. Languages. Does the model list the languages I need?
  2. Features. Voice cloning and streaming are listed per model.
  3. Pricing tiers and license. These are quoted differently by every vendor, so I read them on the model page.
  4. Sources. Each model page links to where its details came from, so I can confirm them with the vendor.
  5. Official audio samples. I play them, then run my own text.

A caveat

BestTTS has not run an independent listening or latency benchmark, and its compare table marks models that are not tested. Treat the directory as a way to build a shortlist, and make the final choice with your own text and your own ears.

Directory: https://besttts.org/

Top comments (0)