What is this video about?
In this Franklin AI video, published on November 8, 2026, the presenter demonstrates Open Mic, a voice assistant he describes as open source and local, and explains why he uses EdgeRock 4B with it. The walkthrough moves from voice input to the model's white paper and finishes with screen capture and rewriting demos. The video describes the workflow as offline and shows how it fits into the desktop.
Where the Open Mic demo starts
The opening shows Open Mic in use. At 0:33 the presenter compares it with WhisperFlow and at 1:05 introduces EdgeRock 4B. This first part is the place to start if you want to see how the voice workflow feels before hearing about the model itself. The description links the Open Mic and EdgeRock projects, including the model's Hugging Face page.
What the white paper section covers
The white paper discussion starts at 2:08. The presenter then covers recurrent state and Rock Memory Anchors, followed by a section on reported benchmarks at 4:34. At 5:39 the discussion turns to speed, efficiency and token generation. These chapters explain the design and discuss performance claims; the description does not establish a universal speed ranking across hardware and workloads.
When the desktop demos return
The screen capture and reasoning segment begins at 6:53. At 8:37 the presenter returns to live operating system integration and voice rewriting. Those sections are useful if your main question is what the assistant can do with screen context and spoken input in one workflow. If you are deciding whether to try it, the video is a demonstration of the author's setup. Its chapter list does not identify a cross-platform installation test or a side-by-side latency test on identical hardware.
Primary source: www.youtube.com ↗