Spaces:
Running
on
Zero
Running
on
Zero
Token limit?
#1
by
jujutechnology
- opened
Hi. The model only outputs approximately 16 seconds of audio no matter how long the text is. Is this just a limitation of this Space or is the model not able to do longer form text?
This model is pre-trained on audio up to 15 seconds, which is okay for streaming but not very good for generating long sentences. On longer sentences it may show instability.
is there a streaming implementation example somewhere ? if not maybe high-level instructions on how to about it, thank you for incredible work 👍👍👍👍