Open OmarTariq612 opened 3 months ago
Using the HF 🤗 Optimum and the help of @OmarTariq612, you can use our model via this link on any Chromium-based browser enabling a set of flags. We still haven't figured out how to "Quantize" the model to make it faster but it works on my AMD Radeon R5 M445 with 8 tokens/s average.
whisper-web: https://github.com/xenova/whisper-web/tree/experimental-webgpu
Use this transformer model: https://huggingface.co/omartariq612/whisper-small-augmented-epoch-5