---
language:
- ar
license: apache-2.0
base_model: openai/whisper-medium
tags:
- generated_from_trainer
datasets:
- tarteel-ai/EA-UD
- tarteel-ai/everyayah
metrics:
- wer
model-index:
- name: Whisper Medium ar-quran
  results:
  - task:
      name: Automatic Speech Recognition
      type: automatic-speech-recognition
    dataset:
      name: Common Voice 17.0
      type: tarteel-ai/EA-UD
    metrics:
    - name: Wer
      type: wer
      value: 0.15533980582524273
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->

# Whisper Medium ar-quran

This model is a fine-tuned version of [openai/whisper-medium](https://huggingface.co/openai/whisper-medium) on the Common Voice 17.0 dataset.
It achieves the following results on the evaluation set:
- Loss: 1.0757
- Wer: 0.1553

## Model description

More information needed

## Intended uses & limitations

More information needed

## Training and evaluation data

More information needed

## Training procedure

### Training hyperparameters

The following hyperparameters were used during training:
- learning_rate: 1e-05
- train_batch_size: 16
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.04
- training_steps: 15000

### Training results

| Training Loss | Epoch | Step  | Validation Loss | Wer    |
|:-------------:|:-----:|:-----:|:---------------:|:------:|
| 0.0377        | 1.0   | 15000 | 1.0757          | 0.1553 |


### Framework versions

- Transformers 4.42.0.dev0
- Pytorch 2.3.0+cu121
- Datasets 2.19.1
- Tokenizers 0.19.1

## Citation

```bibtex
@misc{deepdml/whisper-medium-ar-quran-mix,
      title={Fine-tuned Whisper medium ASR model for speech recognition in Arabic},
      author={Jimenez, David},
      howpublished={\url{https://huggingface.co/deepdml/whisper-medium-ar-quran-mix}},
      year={2025}
    }
```