deepdml/whisper-tiny-af-mix-norm

Whisper Tiny af

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

• Loss: 1.3265
• Wer: 49.0563
• Cer: 20.0715

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: 128
• eval_batch_size: 128
• 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: 5000

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
0.0873	0.2	1000	1.1737	43.8442	18.5673
0.0323	1.014	2000	1.2562	46.8571	19.0335
0.0184	1.214	3000	1.2972	42.9957	17.3357
0.0164	2.028	4000	1.3082	45.8528	18.4676
0.0154	2.228	5000	1.3265	49.0563	20.0715

Framework versions

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

Citation

Please cite the model using the following BibTeX entry:

@misc{deepdml/whisper-tiny-af-mix-norm,
      title={Fine-tuned Whisper tiny ASR model for speech recognition in Afrikaans},
      author={Jimenez, David},
      howpublished={\url{https://huggingface.co/deepdml/whisper-tiny-af-mix-norm}},
      year={2026}
    }