---
license: mit
pipeline_tag: text-to-speech
language: [hi, as, bn, gu, mr, kn, ml, or, pa, ta, te]
base_model: ai4bharat/IndicF5
---
# IndicF5 (unmodified mirror for ReelGen)

An unmodified, ungated mirror of [ai4bharat/IndicF5](https://huggingface.co/ai4bharat/IndicF5)
at revision `fb652a96e508091045890dae5bb99fe511facbcf`, redistributed under its MIT license.
All credit belongs to AI4Bharat. Please cite their work:

> IndicF5: High-Quality Text-to-Speech for Indian Languages — AI4Bharat, IIT Madras.

The weights are trained on Rasa, IndicTTS, LIMMITS and IndicVoices-R; see the original model card.

| File | Source | SHA-256 |
|---|---|---|
| `model.safetensors` | ai4bharat/IndicF5 | `ba7f3671180fb7784e24bd1dafc96e729a38ce02e7f6d3877cdef32525a1865c` |
| `checkpoints/vocab.txt` | ai4bharat/IndicF5 | `d3a5ff6aac12ea4fb50628a66a97b20c1e83b9e5ca356e5afae17b511cda96df` |
| `vocos/config.yaml` | [charactr/vocos-mel-24khz](https://huggingface.co/charactr/vocos-mel-24khz) @ `0feb3fdd` (MIT) | `da9033922f969a47f0c160010226919e59f27761fd5066f3828d46de6650b0fc` |