# Tunisian Arabic ASR Model with wav2vec2

This repository provides all the necessary tools to perform automatic speech recognition from an end-to-end system pretrained on Tunisian arabic dialect

## Performance
the performance of the mode is : 
| Release Version | |WER (%) | CER (%) |
|-----------------|----|---------|---------|
| v1.0            | Without LM |11.82    | 6.33   |
## Dataset
This ASR model was trained on :
* TARIC : The corpus, named TARIC (Tunisian Arabic Railway Interaction Corpus) has a collection of audio recordings and transcriptions from dialogues in the Tunisian Railway Transport Network. - [Taric Corpus](https://aclanthology.org/L14-1385/) -
* STAC :A corpus of spoken Tunisian Arabic - [STAC Corpus](https://www.researchgate.net/publication/307583782_Spoken_Tunisian_Arabic_Corpus_STAC_Transcription_and_Annotation)
* IWSLT : A Tunisian conversational speech - [IWSLT Corpus](https://iwslt.org/2022/dialect)-
* Tunspeech : Our custom dataset

## Install
```python
pip install speechbrain transformers
```