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: 0.3560
Wer Raw: 19.6036
Cer Raw: 7.2565
Wer: 19.6036
Cer: 7.2565

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_ratio: 0.04
training_steps: 20000

Training results

Training Loss	Epoch	Step	Validation Loss	Wer Raw	Cer Raw	Wer	Cer
0.5259	0.05	1000	0.5153	27.2204	9.7771	27.1670	9.7663
0.4068	0.1	2000	0.4695	24.9782	9.1500	24.9598	9.1466
0.2546	0.15	3000	0.4376	23.3862	8.6061	23.3767	8.6047
0.2461	0.2	4000	0.4191	22.2918	8.0878	22.2880	8.0871
0.2203	0.25	5000	0.4099	22.2696	8.1756	22.2658	8.1750
0.2441	0.3	6000	0.4001	21.4796	7.8921	21.4790	7.8920
0.2318	0.35	7000	0.3908	21.4504	7.9370	21.4504	7.9370
0.4077	0.4	8000	0.3833	20.7589	7.6047	20.7589	7.6047
0.1844	0.45	9000	0.3808	20.4431	7.5154	20.4431	7.5154
0.2673	0.5	10000	0.3750	20.3490	7.3850	20.3484	7.3849
0.1677	0.55	11000	0.3726	20.3262	7.5745	20.3262	7.5745
0.1542	0.6	12000	0.3705	19.9080	7.2882	19.9080	7.2882
0.1609	0.65	13000	0.3647	19.8158	7.3612	19.8158	7.3612
0.1483	0.7	14000	0.3607	19.8438	7.4117	19.8438	7.4117
0.1343	0.75	15000	0.3607	19.5616	7.2026	19.5616	7.2026
0.1379	1.0116	16000	0.3598	19.5146	7.1944	19.5146	7.1944
0.1271	1.0616	17000	0.3589	19.7453	7.2933	19.7453	7.2933
0.1309	1.1117	18000	0.3558	19.7154	7.3180	19.7154	7.3180
0.1561	1.1617	19000	0.3573	19.5197	7.0518	19.5197	7.0518
0.206	1.2117	20000	0.3560	19.6036	7.2565	19.6036	7.2565

Framework versions

Transformers 4.48.0.dev0
Pytorch 2.5.1+cu121
Datasets 3.6.0
Tokenizers 0.21.0

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}
    }