Open petersabry opened 5 years ago
I was just thinking about this today. That's definitely something I want to implement.
I was planning on deving something like this too(i just really hated deving with ffmpeg). So +1 from me. (or rather +10000)
This feature would save me literally 1 hr per Video as cutting down LP footage where i dont speak all the time is really frustrating especially when the editing software just randomly decides to go "uhm no byebye".
I think using a ML speech detection model or waveform analyzing algo would be pretty useful in this case!
But maybe audio based editing can cut very important scenes from video, I guess it can be used only certain cases.
It could definitely be achieved. The easiest path would be through frequency checking. 100Hz is the human voice so that should be a good starting point.
Why not? It would be better, maybe as a second option to edit the video.
Like skipping silent moments.
Or even comparing the start frame and the ending frame based on the duration and see whether it was changed or not if so you could have option like fast forwarding if there was no audio.
And so on..
Just a note: i didn't read the source code, maybe you're doing that already idk