whisper-small-sr

Fine-tuned OpenAI Whisper Small.

Output script: this model is intended to produce Serbian Latin only.

• WER on Common Voice 24.0 Serbian test: 7.09%

Model description

Training and evaluation data

This model was fine-tuned on a mixture of publicly available Serbian speech corpora, including:

• Mozilla Common Voice 24.0, evaluated on CV test (sr)
• FLEURS Serbian
• ParlaSpeech-RS (subset of the full dataset)
• Additional Serbian corpora used in the training pipeline

Training procedure

• Epochs: 8
• Batch size: 32
• Optimizer: AdamW
• LR: 6e-5 with warmup (50 steps) + cosine decay to min_lr = 1e-7
• Mixed precision: bfloat16
• SpecAugment: frequency + time masking
• Sampling: weighted sampling across datasets

Training results

Epoch	Train loss	CV WER
1	0.331	0.1562
2	0.338	0.1202
3	0.241	0.1062
4	0.187	0.0913
5	0.150	0.0853
6	0.122	0.0745
7	0.106	0.0709

Evaluation Metrics

• WER (normalized) on Common Voice 24.0 Serbian test: 7.09%
• Text normalization used for WER:
  • punctuation removed
  • lowercased
  • Cyrillic → Latin conversion
  • numbers converted to words