Asr Whisper Medium Commonvoice Fa

🚀 whisper medium fine-tuned on CommonVoice-14.0 Farsi

This repository offers all essential tools for automatic speech recognition using an end - to - end whisper model fine - tuned on CommonVoice (Farsi Language) within SpeechBrain. For a better experience, explore SpeechBrain.

✨ Features

• End - to - End ASR: Perform automatic speech recognition with a fine - tuned whisper model.
• Flexible Setup: The model can be trained and used with SpeechBrain, a user - friendly speech toolkit.

📦 Installation

First, install transformers and SpeechBrain using the following command:

pip install speechbrain transformers

If you want to train the model from scratch, follow these additional steps:

1. Clone SpeechBrain:

git clone https://github.com/speechbrain/speechbrain/

2. Install it:

cd speechbrain
pip install -r requirements.txt
pip install -e .

💻 Usage Examples

Basic Usage

To transcribe your own Farsi audio files:

from speechbrain.inference.ASR import WhisperASR

asr_model = WhisperASR.from_hparams(source="speechbrain/asr-whisper-medium-commonvoice-fa", savedir="pretrained_models/asr-whisper-medium-commonvoice-fa")
asr_model.transcribe_file("speechbrain/asr-whisper-medium-commonvoice-fa/example-fa.mp3")

Advanced Usage

To perform inference on the GPU, add run_opts={"device":"cuda"} when calling the from_hparams method:

asr_model = WhisperASR.from_hparams(source="speechbrain/asr-whisper-medium-commonvoice-fa", savedir="pretrained_models/asr-whisper-medium-commonvoice-fa", run_opts={"device":"cuda"})
asr_model.transcribe_file("speechbrain/asr-whisper-medium-commonvoice-fa/example-fa.mp3")

📚 Documentation

Pipeline description

This ASR system consists of whisper encoder - decoder blocks:

• The pretrained whisper - medium encoder is frozen.
• The pretrained Whisper tokenizer is used.
• A pretrained Whisper - medium decoder ([openai/whisper - medium](https://huggingface.co/openai/whisper - medium)) is finetuned on CommonVoice ar. The obtained final acoustic representation is given to the greedy decoder.

The system is trained with recordings sampled at 16kHz (single channel). The code will automatically normalize your audio (i.e., resampling + mono channel selection) when calling transcribe_file if needed.

Training

To train the model from scratch:

cd recipes/CommonVoice/ASR/transformer/
python train_with_whisper.py hparams/train_fa_hf_whisper.yaml --data_folder=your_data_folder

You can find the training results (models, logs, etc) here.

🔧 Technical Details

The performance of the model is as follows:

📄 License

This project is licensed under the apache - 2.0 license.

Additional Information

Limitations

⚠️ Important Note

The SpeechBrain team does not provide any warranty on the performance achieved by this model when used on other datasets.

Referencing SpeechBrain

@misc{SB2021,
    author = {Ravanelli, Mirco and Parcollet, Titouan and Rouhe, Aku and Plantinga, Peter and Rastorgueva, Elena and Lugosch, Loren and Dawalatabad, Nauman and Ju - Chieh, Chou and Heba, Abdel and Grondin, Francois and Aris, William and Liao, Chien - Feng and Cornell, Samuele and Yeh, Sung - Lin and Na, Hwidong and Gao, Yan and Fu, Szu - Wei and Subakan, Cem and De Mori, Renato and Bengio, Yoshua },
    title = {SpeechBrain},
    year = {2021},
    publisher = {GitHub},
    journal = {GitHub repository},
    howpublished = {\\\\url{https://github.com/speechbrain/speechbrain}},
  }

About SpeechBrain

💡 Usage Tip

SpeechBrain is an open - source and all - in - one speech toolkit. It is designed to be simple, extremely flexible, and user - friendly. Competitive or state - of - the - art performance is obtained in various domains.

Website: https://speechbrain.github.io/

GitHub: https://github.com/speechbrain/speechbrain