Whisper Tiny es

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: 0.3211
Wer Raw: 18.2600
Cer Raw: 6.9620
Wer: 18.2600
Cer: 6.9620

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: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 0.04
training_steps: 20000

Training results

Training Loss	Epoch	Step	Validation Loss	Wer Raw	Cer Raw	Wer	Cer
0.3481	0.05	1000	0.5257	28.4368	10.5617	28.3446	10.5433
0.3051	0.1	2000	0.4624	25.3691	9.3246	25.3424	9.3199
0.2919	0.15	3000	0.4321	23.2941	8.3920	23.2814	8.3897
0.3502	0.2	4000	0.4101	23.0875	8.4403	23.0799	8.4391
0.3659	0.25	5000	0.3921	22.7469	8.6860	22.7456	8.6858
0.2285	0.3	6000	0.3766	21.0780	7.9432	21.0774	7.9431
0.3091	0.35	7000	0.3682	21.3030	8.1103	21.3030	8.1103
0.3405	0.4	8000	0.3600	19.9461	7.2819	19.9461	7.2819
0.2460	1.0134	9000	0.3508	19.7345	7.3612	19.7345	7.3612
0.1760	1.0634	10000	0.3442	19.3303	7.0926	19.3303	7.0926
0.1748	1.1134	11000	0.3414	19.5025	7.4110	19.5025	7.4110
0.2497	1.1634	12000	0.3382	19.0271	6.9921	19.0271	6.9921
0.2383	1.2134	13000	0.3337	19.1015	7.1604	19.1015	7.1604
0.1808	1.2634	14000	0.3313	18.9782	7.1947	18.9782	7.1947
0.1933	1.3134	15000	0.3274	18.7723	7.0631	18.7723	7.0631
0.3308	1.3634	16000	0.3264	19.1072	7.4263	19.1072	7.4263
0.2191	1.4134	17000	0.3241	18.2956	6.9179	18.2956	6.9179
0.1883	2.0268	18000	0.3226	18.0402	6.5623	18.0402	6.5623
0.2357	2.0768	19000	0.3217	17.9105	6.6238	17.9105	6.6238
0.1993	2.1268	20000	0.3211	18.2600	6.9620	18.2600	6.9620

Framework versions

Transformers 5.14.1
Pytorch 2.6.0+cu124
Datasets 5.0.1
Tokenizers 0.22.2

Citation

Please cite the model using the following BibTeX entry:

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