Announcement
A Gradio workflow for trying Audio8 ASR Infinite speech recognition
A post links to the workflow and says the streaming model decodes 12.5 times per second.
TLDR
The post describes Audio8 ASR Infinite as a streaming speech-recognition model and says a rolling KV cache keeps memory use and latency constant, even during 24/7 operation. It links to a Gradio workflow for trying the model.
Combined views
16.4K
1 Source, first seen ago
