Spaces:
Running
Running
[MODELS] suggest new models here
#2
by
cdminix
- opened
If you want to test a model that is open-source, please consider reading the submission instructions at the "Submit here!" tab here: https://huggingface.co/spaces/ttsds/benchmark
If you're unable to submit the data yourself, you can request the model in this thread!
Checklist (from Vaibhavs10/open-tts-tracker)
- Amphion VALL-E
- Amphion VALL-E v2
- Amphion NaturalSpeech2
- Bark
- EmotiVoice
- HierSpeech++
- IMS Toucan
- Maha TTS
- GPT-SoVITS
- MetaVoice
- OpenVoice
- OpenVoice v2
- Parler TTS Mini
- Parler TTS Large
- Pheme
- StyleTTS2
- TorToiSe
- VoiceCraft
- Vokan
- WhisperSpeech
- XTTSv2
- SpeechT5
And from TTS-AGI/TTS-Arena suggestions
Others
Excluded due to lack of speaker prompting
- GlowTTS
- Neural HMM
- P-Flow
- Piper
- RAD-MMM
- RAD-TTS
- Silero
- Tacotron2
- TTTS
- MMS-TTS
- MatchaTTS
- xVA-Synth
- Amphion VITS
- 2Noise/ChatTTS
- PeechV2
Due to the number of systems without speaker prompting or multiple speakers, I'm considering a separate leaderboard for these systems.
Just added TorToiSe.
Added HierSpeech++
Added MetaVoice-1B, and the Amphion unofficial implementations of NaturalSpeech2 and VALLE.
Added Fish Speech (very close to top models, but loses out on duration modeling) and Bark (not great overall, maybe not officially supporting voice cloning plays a role).