deepdml/whisper-base-ar-mix-norm

Whisper Base ar

This model is a fine-tuned version of openai/whisper-base on the Common Voice 17.0 dataset. It achieves the following results on the evaluation set:

• Loss: 2.6739
• Wer: 60.7677

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: 64
• 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_steps: 500
• training_steps: 5000

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.6127	0.2	1000	2.1392	51.3435
0.3365	0.4	2000	2.4225	48.5384
0.2029	0.6	3000	2.5954	50.8035
0.1403	0.8	4000	2.7056	53.3236
0.1053	1.0	5000	2.6739	60.7677

Framework versions

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

Citation

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