README.md

---
language:
- dv
metrics:
- wer
- cer
pipeline_tag: automatic-speech-recognition
tags:
  - audio
  - automatic-speech-recognition
license: mit
library_name: ctranslate2
---

# Whisper small-dv model for CTranslate2

This repository contains the conversion of [whisper-small-dv](https://huggingface.co./davidggphy/whisper-small-dv) to the [CTranslate2](https://github.com/OpenNMT/CTranslate2) model format.
The model is a finetuned version of openai/whisper-small to Divehi language using the [Common Voice 13 dataset](https://huggingface.co./datasets/mozilla-foundation/common_voice_13_0)
This model can be used in CTranslate2 or projects based on CTranslate2 such as [faster-whisper](https://github.com/guillaumekln/faster-whisper).

## Example

```python
from faster_whisper import WhisperModel

# load from local folder
# model = WhisperModel("whisper-small-dv-ct2")
# load from the hub
model = WhisperModel("davidggphy/whisper-small-dv-ct2")

segments, info = model.transcribe("audio.mp3")
for segment in segments:
    print("[%.2fs -> %.2fs] %s" % (segment.start, segment.end, segment.text))
```

## Conversion details

The original model was converted with the following command:

```
ct2-transformers-converter --model davidggphy/whisper-small-dv --output_dir whisper-small-dv-ct2 --copy_files tokenizer.json --quantization float16
```

Note that the model weights are saved in FP16. This type can be changed when the model is loaded using the [`compute_type` option in CTranslate2](https://opennmt.net/CTranslate2/quantization.html).

## More information

**For more information about the original model, see its [model card](https://huggingface.co./davidggphy/whisper-small-dv).**
---