Matthew Baas

icon indicating a website link https://rf5.github.io

icon location Western Cape, South Africa icon location Electronic engineer and speech researcher. I'm interested in speech synthesis and generative speech models.

Results 4 comments of Matthew Baas

contrib: makeseeds.py improvements

I've tried to address some of the more minor points of improvement in #24818 , namely differentiating max seeds per ASN for ipv4 and ipv6, with some more docs and...

Question about the Used Hardware

Hi @CCMaure The current code performs default inference using WavLM to obtain the matching and query set features. Being a transformer model, WavLM's memory requirements scales with the **square** of...

[BUG] RuntimeError: Calculated padded input size per channel: (6). Kernel size: (7). Kernel size can't be greater than actual input size

This seems to happen when Mars 5 fails to generate any output audio and predicts an token as the first output. Can you double check that your **prompt/reference** transcript is...

Using a Generator Network Before the Vocoder

Hi @saeedzou That's an interesting idea! It definitely sounds like it should work fairly well, but I don't have a good feeling for the tradeoffs it might have. It might...