MosesJoshuaCoker/Krio_fine_tune_novax

---

language: - kri license: apache-2.0 tags: - whisper - automatic-speech-recognition - krio - sierra-leone datasets: - MosesJoshuaCoker/NovaxDataset metrics: - wer - cer

🎙️ Whisper Small Fine-tuned for Krio (Sierra Leone)

Model Description

This model is a fine-tuned version of openai/whisper-small on the NumbersD dataset for Krio speech recognition.

Performance

• Word Error Rate (WER): 4.27%
• Character Error Rate (CER): 1.54%

Usage

from transformers import WhisperProcessor, WhisperForConditionalGeneration
import librosa

processor = WhisperProcessor.from_pretrained("MosesJoshuaCoker/Krio_fine_tune_novax")
model = WhisperForConditionalGeneration.from_pretrained("MosesJoshuaCoker/Krio_fine_tune_novax")

# Load Krio audio
audio, sr = librosa.load("your_krio_audio.wav", sr=16000)
inputs = processor(audio, sampling_rate=16000, return_tensors="pt")
predicted_ids = model.generate(inputs.input_features)
transcription = processor.batch_decode(predicted_ids, skip_special_tokens=True)[0]
print(f"Krio: {transcription}")

### **2. Test Your Model**
You can now test it directly from any Python environment:

```python
# Test that it works
from transformers import pipeline
pipe = pipeline("automatic-speech-recognition", model="MosesJoshuaCoker/Krio_fine_tune_novax")
result = pipe("https://huggingface.co/datasets/MosesJoshuaCoker/NumbersD/resolve/main/audio/001.wav")
print(result["text"])