Zanshin
Navigate through media by speaker
5 followers
Navigate through media by speaker
5 followers
In YouTube video and your own files: visualize who speaks when & for how long, jump/skip speaker segments, set different speeds for each speaker, and auto-skip speakers
残心 / Zanshin is a better way to listen to podcasts, interviews, press conferences, etc.
I built this after being frustrated listening to podcasts/interviews where I only wanted to listen to certain people on there, but had to manually scrub through to find them. Or I wanted to quickly & accurately skip to the next speaker if someone started rambling or started talking about something I wasn't interested in. Also, in some podcasts, some people talk way slower than others, so I wanted the ability to selectively speed them up.
Download for macOS here: https://zanshin.sh
Windows & Linux also supported, but you'll have to run some terminal commands (for now).
See these instructions: https://zanshin.sh/dev_instructions
For the technical crowd: Zanshin is made possible by a very fast speaker diarization pipeline I've developed called Senko. It's much faster than any other open source alternative I found, and is what enables Zanshin to run efficiently and fast on a consumer laptop like a MacBook. It runs even faster on NVIDIA.
Check out Senko here: https://senko.sh
Cheers, everyone. I hope you find 残心 / Zanshin useful.
Different speeds per speaker is genius for podcasts/interviews. Does it work with live streams or just uploaded videos?
Also launching today (ZenTrack) - upvoted!
@graino Yeah I find it really useful!
Only works with uploaded videos, not livestreams, unfortunately, as that would require real-time diarization.
ZenTrack looks great! I definitely have trouble staying focussed, so looks very useful.
Cheers
@hamza_q_ Thanks Hamza! Makes sense about livestreams needing real-time processing. Really appreciate you checking out ZenTrack and the kind words. Best of luck with Zanshin!
@graino Thanks! And likewise.