You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Request to prioritize releasing the 25Hz tokenizer and models
#391
Could you prioritize the open-source release of the 25Hz tokenizer and Qwen3-TTS models described in the technical report?
My use case is multilingual dubbing for short dramas, where speaker consistency and expressive delivery are important. I would like to evaluate the 25Hz models alongside the released 12Hz models. I am particularly interested in the 25Hz VoiceEditing model, which is described as supporting both voice cloning and instruction following, as well as the 25Hz models' reported long-form generation performance.
The following would be especially helpful:
25Hz tokenizer weights and code, including audio encoding and waveform decoding.
25Hz TTS model weights, especially Base and VoiceEditing.
Runnable inference examples for voice cloning and instruction-controlled generation.
If a release is not imminent, could you share an approximate timeline and the planned release scope? A staged release would also be useful for developers who want to begin evaluating the 25Hz approach.
Thank you for your work on Qwen3-TTS!
How important is releasing the 25Hz tokenizer and models for your use case?
High priority — I need the 25Hz models for evaluation or development.
100%
Interested — a release roadmap would be sufficient for now.
0%
Low priority — the released 12Hz models meet my current needs.
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
Uh oh!
There was an error while loading. Please reload this page.
Hi Qwen team,
Could you prioritize the open-source release of the 25Hz tokenizer and Qwen3-TTS models described in the technical report?
My use case is multilingual dubbing for short dramas, where speaker consistency and expressive delivery are important. I would like to evaluate the 25Hz models alongside the released 12Hz models. I am particularly interested in the 25Hz VoiceEditing model, which is described as supporting both voice cloning and instruction following, as well as the 25Hz models' reported long-form generation performance.
The following would be especially helpful:
If a release is not imminent, could you share an approximate timeline and the planned release scope? A staged release would also be useful for developers who want to begin evaluating the 25Hz approach.
Thank you for your work on Qwen3-TTS!
1 vote ·
All reactions