---
pipeline_tag: automatic-speech-recognition
language: mri
license: apache-2.0
tags:
  - trimmed
library_name: transformers
base_model: openai/whisper-large-v3-turbo
base_model_relation: quantized
datasets:
  - lbourdois/fineweb-2-trimming
---

# whisper-large-v3-turbo-mri-32768
This model is a **3.02% whisper-large-v3-turbo** version of [openai/whisper-large-v3-turbo](https://huggingface.co/openai/whisper-large-v3-turbo) optimized for **Maori** language via vocabulary size reduction using the [trimming](https://huggingface.co/blog/lbourdois/introduction-to-trimming) method.  
This trimmed model should perform similarly to the original model with only 32,768 tokens and a much smaller memory footprint. However, it may not perform well for other languages as tokens not commonly used in the selected languages were removed from the vocabulary.

## Model Statistics
| Metric | Original | Trimmed | Reduction |
|--------|----------|---------|-----------|
| **Vocabulary size** | 51,865 tokens | 32,768 tokens | **36.82%** |
| **Model size** | 808,878,080 params | 784,432,640 params | **3.02%** |

![image](https://raw.githubusercontent.com/lbourdois/blog/refs/heads/master/assets/images/Trimming/whisper-large-v3-turbo-32768.png)

## Mining Dataset Statistics
- **Number of texts used for mining**: 158,804 texts  
- **Dataset**: [lbourdois/fineweb-2-trimming](https://huggingface.co/datasets/lbourdois/fineweb-2-trimming)

## Usage
```python
from transformers import AutoModelForSpeechSeq2Seq, AutoProcessor, pipeline
import librosa

# Pipeline function
processor = AutoProcessor.from_pretrained("alphaedge-ai/whisper-large-v3-turbo-mri-32768")
pipe = pipeline(
    "automatic-speech-recognition",
    model="alphaedge-ai/whisper-large-v3-turbo-mri-32768",
    tokenizer=processor.tokenizer,
    feature_extractor=processor.feature_extractor,
    generate_kwargs={"language": "maori", "task": "transcribe"},
)

# Loading and resampling at 16 kHz (required by Whisper)
audio_array, sampling_rate = librosa.load(audio_path, sr=16000)

# Result
result = pipe(audio_array)
print("Transcription :", result["text"])
```

## Citations

#### Whisper
```
@misc{radford2022whisper,
  doi = {10.48550/ARXIV.2212.04356},
  url = {https://arxiv.org/abs/2212.04356},
  author = {Radford, Alec and Kim, Jong Wook and Xu, Tao and Brockman, Greg and McLeavey, Christine and Sutskever, Ilya},
  title = {Robust Speech Recognition via Large-Scale Weak Supervision},
  publisher = {arXiv},
  year = {2022},
  copyright = {arXiv.org perpetual, non-exclusive license}
}
```

#### Trimming blog post
```
@misc{hf_blogpost_trimming,
      title={Introduction to Trimming}, 
      author={Loïck BOURDOIS and Tom AARSEN and Bram VANROY and Christopher AKIKI and Woojun JUNG and Manuel ROMERO and Prithiv SAKTHI},
      year={2026},
      url={https://huggingface.co/blog/lbourdois/introduction-to-trimming}, 
}
```