Asr Wav2vec2 Transformer Aishell

🚀 Transformer for AISHELL + wav2vec2 (Mandarin Chinese)

This repository offers all the essential tools for automatic speech recognition. It utilizes an end - to - end system pretrained on AISHELL + wav2vec2 (Mandarin Chinese) within SpeechBrain. For an enhanced experience, we suggest you explore more about SpeechBrain.

The model's performance is presented as follows:

🚀 Quick Start

This ASR system consists of 2 distinct yet linked components:

• A Tokenizer (unigram) that converts words into subword units and is trained with the train transcriptions of LibriSpeech.
• An Acoustic model composed of a wav2vec2 encoder and a joint decoder with CTC + transformer. Thus, the decoding also integrates the CTC probabilities.

To train this system from scratch, refer to our SpeechBrain recipe.

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

✨ Features

• Utilizes an end - to - end system pretrained on AISHELL + wav2vec2 for Mandarin Chinese speech recognition.
• Incorporates a unigram tokenizer and a wav2vec2 - based acoustic model with CTC + transformer decoding.
• Automatically normalizes audio during transcription.

📦 Installation

First, install SpeechBrain using the following command:

pip install speechbrain

We recommend that you read our tutorials and learn more about SpeechBrain.

💻 Usage Examples

Basic Usage

Transcribing your own audio files (in English):

from speechbrain.inference.ASR import EncoderDecoderASR
asr_model = EncoderDecoderASR.from_hparams(source="speechbrain/asr-wav2vec2-transformer-aishell", savedir="pretrained_models/asr-wav2vec2-transformer-aishell")
asr_model.transcribe_file("speechbrain/asr-wav2vec2-transformer-aishell/example_mandarin.wav")

Advanced Usage

Inference on GPU

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

Parallel Inference on a Batch

To learn how to transcribe a batch of input sentences in parallel using a pre - trained model, check this Colab notebook.

Training

The model was trained with SpeechBrain (Commit hash: '480dde87'). To train it from scratch, follow these steps:

1. Clone SpeechBrain:

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

2. Install it:

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

3. Run Training:

cd recipes/AISHELL-1/ASR/transformer/
python train.py hparams/train_ASR_transformer_with_wav2vect.yaml --data_folder=your_data_folder

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

Limitations

The SpeechBrain team does not provide any warranty for the model's performance when used on other datasets.

📚 Documentation

About SpeechBrain

• Website: https://speechbrain.github.io/
• Code: https://github.com/speechbrain/speechbrain/
• HuggingFace: https://huggingface.co/speechbrain/

Citing SpeechBrain

Please cite SpeechBrain if you use it for your research or business.

@misc{speechbrain,
  title={{SpeechBrain}: A General-Purpose Speech Toolkit},
  author={Mirco Ravanelli and Titouan Parcollet and Peter Plantinga and Aku Rouhe and Samuele Cornell and Loren Lugosch and Cem Subakan and Nauman Dawalatabad and Abdelwahab Heba and Jianyuan Zhong and Ju-Chieh Chou and Sung-Lin Yeh and Szu-Wei Fu and Chien-Feng Liao and Elena Rastorgueva and François Grondin and William Aris and Hwidong Na and Yan Gao and Renato De Mori and Yoshua Bengio},
  year={2021},
  eprint={2106.04624},
  archivePrefix={arXiv},
  primaryClass={eess.AS},
  note={arXiv:2106.04624}
}

📄 License

This project is licensed under the "apache - 2.0" license.