JONAHKYAGABA/Language-Model-Integration-with-Wav2Vec2-for-Oromo-and-Amharic

Language Model Integration with Wav2Vec2 for Oromo and Amharic # 🧠 Language Model Integration with Wav2Vec2 for Oromo and Amharic This project boosts **Wav2Vec2** speech recognition models using a **KenLM 5-gram language model**, enhancing decoding quality for **low-resource African languages** such as Oromo and Amharic. --- ## 🎯 Objectives - Download or prepare `.arpa` language models. - Build a binary KenLM (`.bin`) for fast decoding. - Use `pyctcdecode` to attach the LM to Wav2Vec2. - Save the enhanced `Wav2Vec2ProcessorWithLM`. - Upload the processor to Hugging Face with Git LFS. --- ## 🧰 Dependencies Install required libraries: ```bash pip install datasets transformers torchaudio librosa jiwer evaluate pyctcdecode \ huggingface_hub wandb matplotlib seaborn accelerate # KenLM and build tools sudo apt install build-essential cmake libboost-all-dev zlib1g-dev libbz2-dev liblzma-dev git-lfs