The inference speed is very slow

rednote-hilab/dots.tts

Status: Open

Opened: Jun 9, 2026

Comments: 7

The inference speed is very slow, with an RTF of more than 2. I am using an L40 GPU, and I would like to know what the benchmark RTF is for model testing.

Python

View on GitHub ↗

Other Comments / Reviews

Thanks for the suggestion, but currently we don't have pl...

by xlians555 Jun 9, 2026
Do you have plans to make this work in pure C++, like thi...

by GalenMarek14 Jun 9, 2026
```python from dots_tts.runtime import DotsTtsRuntime ...

by xlians555 Jun 9, 2026
@xlians555 Is there any example of `generate_stream` ?

by ukemamaster Jun 9, 2026
You can add the `--optimize` flag in current PyTorch vers...

by xlians555 Jun 9, 2026