4 ms·
How is this so much faster than even GPU-based whisper?
by ranger_danger 5mo ago
How is this so much faster than even GPU-based whisper?
- mrkn1 5mo agosmall, ONNX-optimized models designed specifically for low-latency CPU streaming, so it avoids overhead of large transformer arch and GPU memory transfers