Model reference · open weights
F5-Hindi-24KHz is an open-weight audio or speech model from SPRINGLab. F5-Hindi-24KHz (BF16) weighs 632 MB; the smallest configuration that runs it is RTX 3060 12 GB.
For self-hosting, compare this build's requirements below with the GPU you plan to rent.
Summary of the SPRINGLab/F5-Hindi-24KHz model card, 2026-10-01
What it is
| Released by | SPRINGLab |
|---|---|
| Released | 2024-11-27 |
| VRAM | 632 MB for the weights |
What it runs on
| Card | One stream | Memory |
|---|---|---|
| RTX 3060 12 GB | fits | 11.6 GB |
| RTX 4060 Ti 16 GB | fits | 15.4 GB |
| RTX 3090 24 GB | fits | 23.4 GB |
| RTX 4090 24 GB | fits | 23.4 GB |
| RTX 5090 32 GB | fits | 31.0 GB |
| L40S 48 GB | fits | 44.0 GB |
| A100 80 GB | fits | 78.2 GB |
| H100 80 GB | fits | 78.1 GB |
| RTX PRO 6000 Blackwell 96 GB | fits | 93.8 GB |
| DGX Spark (GB10) 128 GB unified | fits | 107 GB |
| H200 141 GB | fits | 138 GB |
| B200 180 GB | fits | 176 GB |
From the model card
This is a Hindi Text-to-Speech model trained from scratch using the F5 architecture.
The model was developed and is primarily intended for research purposes.
Clone the following github repo and refer to the README: https://github.com/rumourscape/F5-TTS
The model was trained on 8x A100 40GB GPUs for close to a week. We would like to thank CDAC for providing the compute resources.
We used the "small" configuration(151M parameter) model for training according to the F5 paper.
We used the Hindi subsets of IndicTTS and IndicVoices-R datasets for training this model.
Quoted from the model card on Hugging Face. The full card is behind the Hugging Face link above.
How it works