Whisper Tamil Small

🚀 Whisper Tamil Small

This model is a fine - tuned version of openai/whisper-small on Tamil data from multiple publicly available ASR corpuses. It was fine - tuned as part of the Whisper fine - tuning sprint, aiming to enhance the performance of automatic speech recognition in the Tamil language.

NOTE: The code for training this model can be reused from the whisper-finetune repository.

🚀 Quick Start

To evaluate this model on an entire dataset, use the evaluation codes in the whisper-finetune repository. The same repository also offers scripts for faster inference with whisper - jax.

✨ Features

• Fine - tuned on multiple Tamil ASR corpuses for better performance in Tamil speech recognition.
• Reusable training code available in a public repository.
• Support for faster inference using whisper - jax.

📦 Installation

No specific installation steps are provided in the original document.

💻 Usage Examples

Basic Usage

To infer a single audio file using this model, use the following code snippet:

>>> import torch
>>> from transformers import pipeline

>>> # path to the audio file to be transcribed
>>> audio = "/path/to/audio.format"
>>> device = "cuda:0" if torch.cuda.is_available() else "cpu"

>>> transcribe = pipeline(task="automatic-speech-recognition", model="vasista22/whisper-tamil-small", chunk_length_s=30, device=device)
>>> transcribe.model.config.forced_decoder_ids = transcribe.tokenizer.get_decoder_prompt_ids(language="ta", task="transcribe")

>>> print('Transcription: ', transcribe(audio)["text"])

Advanced Usage

For faster inference of whisper models, use the whisper-jax library. Follow the necessary installation steps as mentioned here before using the following code:

>>> import jax.numpy as jnp
>>> from whisper_jax import FlaxWhisperForConditionalGeneration, FlaxWhisperPipline

>>> # path to the audio file to be transcribed
>>> audio = "/path/to/audio.format"

>>> transcribe = FlaxWhisperPipline("vasista22/whisper-tamil-small", batch_size=16)
>>> transcribe.model.config.forced_decoder_ids = transcribe.tokenizer.get_decoder_prompt_ids(language="ta", task="transcribe")

>>> print('Transcription: ', transcribe(audio)["text"])

📚 Documentation

Training and evaluation data

Training Data

• IISc-MILE Tamil ASR Corpus
• ULCA ASR Corpus
• Shrutilipi ASR Corpus
• Microsoft Speech Corpus (Indian Languages)
• Google/Fleurs Train+Dev set
• Babel ASR Corpus

Evaluation Data

• Microsoft Speech Corpus (Indian Languages) Test Set
• Google/Fleurs Test Set
• IISc-MILE Test Set
• Babel Test Set

Training hyperparameters

The following hyperparameters were used during training:

🔧 Technical Details

The model is a fine - tuned version of openai/whisper-small on Tamil data. The fine - tuning was part of the Whisper fine - tuning sprint. The code for training and evaluation is available in the whisper-finetune repository.

📄 License

This model is licensed under the Apache 2.0 license.

Acknowledgement

This work was done at Speech Lab, IIT Madras. The compute resources were funded by the "Bhashini: National Language translation Mission" project of the Ministry of Electronics and Information Technology (MeitY), Government of India.

Model Index