Whisper Small for Quran Recognition

This model is a fine-tuned version of openai/whisper-small on the Quran_Reciters dataset. It achieves the following results on the evaluation set:

  • Loss: 0.0179
  • Wer: 3.1288

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_steps: 500
  • training_steps: 4000
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer
0.0073 1.62 500 0.0249 5.0026
0.0014 3.24 1000 0.0214 4.1086
0.0008 4.85 1500 0.0221 3.9883
0.0 6.47 2000 0.0180 2.9740
0.0 8.09 2500 0.0177 3.0944
0.0 9.71 3000 0.0178 3.0944
0.0 11.33 3500 0.0179 3.1288
0.0 12.94 4000 0.0179 3.1288

Framework versions

  • Transformers 4.40.0.dev0
  • Pytorch 2.1.2
  • Datasets 2.17.1
  • Tokenizers 0.15.1
Downloads last month
25
Safetensors
Model size
242M params
Tensor type
F32
·
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Model tree for AsemBadr/whisper-small-final

Finetuned
(2102)
this model

Evaluation results