---
language:
- lus
license: apache-2.0
pipeline_tag: automatic-speech-recognition
base_model: openai/whisper-large-v3-turbo
tags:
- generated_from_trainer
datasets:
- andrewbawitlung/MiZonal-v3.0
metrics:
- wer
- cer
model-index:
- name: whisper-large-v3-turbo-mizonal3-E1-lus-v2026.06
  results:
  - task:
      name: Automatic Speech Recognition
      type: automatic-speech-recognition
    dataset:
      name: MiZonal v3.0
      type: andrewbawitlung/MiZonal-v3.0
      config: default
      split: test
    metrics:
    - name: Wer
      type: wer
      value: 16.0499
    - name: Cer
      type: cer
      value: 3.1565
    - name: Real Time Factor
      type: rtf
      value: 0.0547
---
![Mizo Automatic Speech Recognition (ASR) Models v3.0](banner.jpg)

# whisper-large-v3-turbo-mizonal3-E1-lus-v2026.06

This model is a fine-tuned version of [openai/whisper-large-v3-turbo](https://huggingface.co/openai/whisper-large-v3-turbo) on the **MiZonal v3.0** dataset.

It achieves the following results on the evaluation set:
- Wer: 16.0499
- Cer: 3.1565
- Real Time Factor: 0.0547

## Quick Inference


```python
import torch
import librosa
from transformers import WhisperProcessor, WhisperForConditionalGeneration

device = "cuda" if torch.cuda.is_available() else "cpu"

processor = WhisperProcessor.from_pretrained("andrewbawitlung/whisper-large-v3-turbo-mizonal3-E1-lus-v2026.06")
model = WhisperForConditionalGeneration.from_pretrained("andrewbawitlung/whisper-large-v3-turbo-mizonal3-E1-lus-v2026.06").to(device)

audio, sr = librosa.load("your_audio.wav", sr=16000)
input_features = processor(audio, sampling_rate=16000, return_tensors="pt").input_features.to(device)

with torch.no_grad():
    predicted_ids = model.generate(input_features, max_new_tokens=256)

transcription = processor.batch_decode(predicted_ids, skip_special_tokens=True)[0]
print(transcription)
```


## Model description

### Experiment Configurations
This repository is part of a series of experiments. The different configurations are:
- **E1 (Baseline):** Standard training configuration.
- **E2 (Noise):** Training with background noise augmentation.
- **E3 (Speed):** Training with speed perturbation augmentation.
- **E4 (SpecAug):** Training with SpecAugment (time and frequency masking).
- **E5 (Combined):** Training with a combination of all augmentations.

### All Models in this Family
| Experiment | Hugging Face Repository |
| :--- | :--- |
| **E1 (Baseline)** | [andrewbawitlung/whisper-large-v3-turbo-mizonal3-E1-lus-v2026.06](https://huggingface.co/andrewbawitlung/whisper-large-v3-turbo-mizonal3-E1-lus-v2026.06) |
| **E2 (Noise)** | [andrewbawitlung/whisper-large-v3-turbo-mizonal3-E2-lus-v2026.06](https://huggingface.co/andrewbawitlung/whisper-large-v3-turbo-mizonal3-E2-lus-v2026.06) |
| **E3 (Speed)** | [andrewbawitlung/whisper-large-v3-turbo-mizonal3-E3-lus-v2026.06](https://huggingface.co/andrewbawitlung/whisper-large-v3-turbo-mizonal3-E3-lus-v2026.06) |
| **E4 (SpecAug)** | [andrewbawitlung/whisper-large-v3-turbo-mizonal3-E4-lus-v2026.06](https://huggingface.co/andrewbawitlung/whisper-large-v3-turbo-mizonal3-E4-lus-v2026.06) |
| **E5 (Combined)** | [andrewbawitlung/whisper-large-v3-turbo-mizonal3-E5-lus-v2026.06](https://huggingface.co/andrewbawitlung/whisper-large-v3-turbo-mizonal3-E5-lus-v2026.06) |

### Training hyperparameters

The following hyperparameters were used during training:
- learning_rate: 1e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: OptimizerNames.ADAMW_TORCH_FUSED
- lr_scheduler_type: SchedulerType.LINEAR
- num_epochs: 8

### Training results

| step | epoch | train_loss | eval_loss | eval_wer | eval_cer | learning_rate | grad_norm |
| --- | --- | --- | --- | --- | --- | --- | --- |
| 250 | 0.9107 | 0.4605 | 0.4389 | 29.9406 | 6.7554 | 4.98e-06 | 9.7224 |
| 500 | 1.8197 | 0.2943 | 0.3240 | 22.6136 | 5.0148 | 9.98e-06 | 7.8017 |
| 750 | 2.7286 | 0.1915 | 0.2703 | 19.6945 | 4.1455 | 8.54e-06 | 6.2710 |
| 1000 | 3.6375 | 0.1138 | 0.2588 | 17.1451 | 3.6383 | 7.06e-06 | 3.8060 |
| 1250 | 4.5464 | 0.0663 | 0.2459 | 16.5807 | 3.4139 | 5.59e-06 | 3.0444 |
| 1500 | 5.4554 | 0.0349 | 0.2604 | 17.6219 | 3.6313 | 4.12e-06 | 3.4490 |
| 1750 | 6.3643 | 0.0175 | 0.2644 | 15.5687 | 3.3150 | 2.65e-06 | 3.3370 |
| 2000 | 7.2732 | 0.0067 | 0.2698 | 15.1990 | 3.2125 | 1.18e-06 | 1.0645 |