[gagan2020]

It is not good for text to speech (TTS) as well. I am trying it for few days. First of all 1.5B model documentation is not there. 0.5B realtime is shit model. I was converting text, line by line and it was randomly adding music and couldn't handle special characters like "…".

I really disappointed with this model to say the least.

[SyneRyder]

> ...it was randomly adding music...

I've been noticing this with the Mistral Voxtral TTS models too. I have my AI record a morning briefing podcast for myself, and occasionally there are sounds like music at the start (the british voice had a musical tone underneath that sounded a little like the end of the BBC News theme). I don't think I've ever encountered that with the OpenAI TTS models, so they're now my default go-to again.

[tjungblut]

yep, it seems this was trained on large amount of podcasts with ad jingles or phone call queues with elevator music. I was also pretty disappointed to run the TTS last week.

[Stagnant]

The 7B parameter Vibevoice TTS model is still the most impressive local TTS model i've tried. It was pulled by Microsoft a few days after its release due to "abuse potential" but it can be found in various community maintained huggingface repos.