iamgarvit/whisper-small-hi-asr

Whisper Small Hindi ASR

This model is a fine-tuned version of openai/whisper-small on the Common Voice Hindi + Hindi Audio Speech dataset. It achieves the following results on the evaluation set:

• Loss: 0.3006
• Wer: 24.9006

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
0.0611	3.0581	1000	0.2043	26.4482
0.0061	6.1162	2000	0.2492	25.2637
0.0006	9.1743	3000	0.2867	24.9611
0.0002	12.2324	4000	0.3006	24.9006

Framework versions

• Transformers 4.40.2
• Pytorch 2.6.0+cu124
• Datasets 2.18.0
• Tokenizers 0.19.1