AI4BharatIndic Parler-TTS
Indic Parler-TTS Pretrained
The pretrained Indic Parler release precedes the more focused fine-tuned model. It was trained on 8,385 hours of Indian-language and English speech.
Restricted access
A larger pretraining corpus serves a different stage
Its prompt tokenizer expands the vocabulary and supports byte fallback for multilingual text. The release remains separate from Indic Parler-TTS because the latter underwent additional fine-tuning.Indic Parler-TTS Pretrained official model card
Use it as a training starting point
Practical guidance. The earlier checkpoint is useful when you want to study the effect of later speech fine-tuning. Use the same transcripts and voice descriptions for that comparison.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Status | Restricted access |
| Access | Downloadable |
| Model identifiers | indic-parler-tts-pretrained |
| Coverage | English and 20 Indian languages.Indic Parler-TTS Pretrained official model card |
| Collaboration | AI4Bharat and the Hugging Face audio team.Indic Parler-TTS Pretrained official model card |
| License | Apache 2.0; gated download.Indic Parler-TTS Pretrained official model card |
Related models
- Indic Parler-TTS
AI4Bharat ยท Indic Parler-TTS
Joint AI4Bharat and Hugging Face speech model controlled through written voice descriptions.
Launch date not confirmedRestricted accessDownloadable