Speech generation · Opt-in
IndexTTS 2.5
Multilingual cloning with emotion control from a reference clip, a vector or text. Made for dubbing.
- BilibiliBy
- 5Languages
- 24 kHzOutput
Can do
- Voice cloningYes. YesSpeaks in a voice copied from a short reference clip.
- Voice designNo. NoBuilds a new voice from a written description.
- Emotion controlYes. YesTakes graded emotion from a clip, a vector or text.
- Preset voicesNo. NoShips fixed voices; no reference clip needed.
Languages
5languages
- Arabic
ar - Chinese
zh - English
en - Japanese
ja - Spanish
es
Runs on
PyTorch, isolated
isolatedPyTorchCUDACPUEngine doc
Hardware
| Engine | CUDA | ROCm | Intel Arc | Apple Silicon | Vulkan | NPU | CPU |
|---|---|---|---|---|---|---|---|
| PyTorch, isolated | Supported | Not supported | Not supported | Not supported | Not supported | Not supported | Supported |
Supported Needs a build Experimental
Licence
- Code
- bilibili Model Use License
- Weights
- bilibili Model Use License; a separate licence is needed above 100M monthly users or RMB 1B annual revenue
VoiceStudio is AGPL-3.0 and doesn't relicense model weights. Licence guide
Related
From the VoiceStudio app at 0834c8b. 27 models listed.