Talos model files
The speech files Talos downloads when it is set up: the speech recognizer and the expressive voice pack. Talos downloads them itself, with your own Hugging Face token, after your access request is approved. The chat and vision models are not here: Talos downloads those from their own publishers' repositories.
What is in this repository
| File | What it is | Licence |
|---|---|---|
| Apex speech recognizer export | Hindi, Hinglish and English speech recognition, converted from Oriserve/Whisper-Hindi2Hinglish-Apex | Apache License 2.0 |
chatterbox-pack.tar.gz | The Chatterbox multilingual voice model from Resemble AI, stored at reduced precision, and 12 voice embeddings of the repository owner's own voice | MIT (the model); the owner's own (the voices) |
chatterbox-venv.tar.gz | The portable Python environment the voice model runs in | Each package under its own licence |
Sources, changes and every licence are listed in THIRD_PARTY_NOTICES.md.
Terms of use
- Use these files only with Talos, for personal use.
- Do not use Talos, or these files, to clone or imitate anyone's voice without their consent. The 12 voices are the repository owner's own, recorded with consent. Generated audio carries Resemble AI's Perth watermark; do not remove it.
- Each file keeps its own licence, as listed in THIRD_PARTY_NOTICES.md.
No warranty
These files are provided as they are, without warranty of any kind, express or implied. The repository owner is not liable for any claim, damage or other liability arising from their use.