No efficient TTS ability from OpenAI?

API Feedback
1

I have been reviewing OpenAI offerings when it comes to TTS.

  • The whisper based TTS speech API is very slow and don’t have a true streaming capability
  • The new offerings in terms of realtime-api or audio-preview api are way too expensive.

What is the recommendation to use if my use case is simply Text to speech?

2

Coqui-tts that is what i use for kruel.ai offline voice system. I am still experimenting with different models to see, some are really fast but not realistic some are not bad, and still fast, others are really slow haha.

you can also look at Edgetts, which is pretty fast.