KBLab/wav2vec2-large-voxrex-swedish
🧠 AI ModelKBLab
Swedish ASR model fine-tuned on VoxRex, achieving high accuracy on Common Voice.
This model is based on Facebook's wav2vec2-large and fine-tuned specifically for Swedish automatic speech recognition using the VoxRex dataset. It leverages the transformer architecture with PyTorch and safetensors for efficient inference. The model is optimized for Swedish phonetic nuances and achieves high word error rate improvements on the Common Voice benchmark. Key features include open-source availability, compatibility with the HuggingFace transformers library, and support for automatic speech recognition pipelines. The model has been downloaded over 1.5 million times and is suitable for production use cases such as transcription, voice assistants, and accessibility tools for Swedish speech.
💡Highlights
- ├─Wav2Vec2 large architecture
- ├─Fine-tuned on VoxRex dataset
- └─1.5M+ downloads on HF
🎯For
- ├─Swedish NLP researchers
- ├─Speech-to-text developers
- └─Language technology companies