finetuned-w2v2-bert-burmese-asr
This model is a fine-tuned version of facebook/w2v-bert-2.0 on the OpenSLR-80 Burmese Speech Dataset dataset. It achieves the following results on the evaluation set:
- Loss: 0.3886
- Wer: 0.4256
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: 5e-05
- train_batch_size: 16
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- optimizer: Use OptimizerNames.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: 500
- num_epochs: 10
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.2901 | 4.1678 | 300 | 0.4585 | 0.5053 |
| 0.2127 | 8.3357 | 600 | 0.3886 | 0.4256 |
🔊 Real Example Evaluation
🎤 Audio Sample
You can listen to an example where a user says:
မင်္ဂလာ ပါ တွေ့ရတာဝမ်းသာပါတယ် ရှင်
📝 Model Transcription
မင်္ဂလာ ပါ တွေ့ရတာ ဝမ်းသပါတယ် ရှင်
✅ Reference
မင်္ဂလာ ပါ တွေ့ရတာဝမ်းသာပါတယ် ရှင်
📉 Word Error Rate (WER)
WER ≈ 0.10 – minor vowel difference only, excellent semantic preservation.
🎧 Audio Playback
Download and listen: burmese_test.wav
Framework versions
- Transformers 4.51.3
- Pytorch 2.6.0+cu124
- Datasets 3.6.0
- Tokenizers 0.21.1