nelovoice/nelo-whisper-darja-speech-model

# nelo-whisper-darja-speech-model Fine-tuned Whisper model for Moroccan Darija speech recognition. ## Contents - 🏋️‍♂️ `finetune_whisper_darija.ipynb`: Notebook for training the Whisper model on Moroccan Darija audio data using batch training. - 🔁 `whisper_darja_inference_cleaned_2epochs.ipynb`: Notebook showing inference results after fine-tuning for 2 epochs. ## Overview This repository focuses on fine-tuning OpenAI’s Whisper model for speech-to-text transcription in Moroccan Darija. It includes training and evaluation notebooks, leveraging a custom cleaned dataset. ## Goals - Adapt Whisper to better understand and transcribe Moroccan Darija. - Compare performance before and after fine-tuning. - Provide reproducible notebooks for training and inference.