SponsorBlock
SponsorBlock copied to clipboard
ajayyy
New Category: Loud Noises
By it's nature, ear-rape clips are obnoxiously loud. Some earrape is unexpected, and as such it risks permanently damaging speakers/headphones if the volume is too high, and even damaging the user's listening capabilities. For example: https://www.youtube.com/watch?v=4Grq-Gz5dh0?t=35s
I think it's important that we mute these segments by default. As for the color, I think it should have a similar color to that of the "Music: Non-Music Section" category, to keep the color coding somewhat consistent.
This is something that can be detected automatically https://github.com/NDevTK/AutoSkip/blob/master/ContentScript.js
Would it be possible to lower the volume to a percentage user set when this segment is detected? Say, on section marked as loud to dip audio to 30%.
edit Oh. #1086
For privacy and offline usage I think this would be better as a OS level script. If such a tool does not already exist I can try to make one if wanted.
@NDevTK I frankly doubt this can be done automatically, as rarely the sounds are just loud, usually they are "jokes" such as amplified and clipped sounds, distortions, single frequency buzzes or beeps or sudden variations or known sounds to provoke reactions, without it actually having a peak in volume.
I already have a compressor installed that does this to all Windows audio after getting acquainted to some very funny streamers, but effect is limited. The reason for this is that simple loudness doesn't describe discomfort. 80% volume nail on chalkboard is way worse than 100% pure sine 50 Hz.
If you can crack this, you'll be famous. But I'm afraid that nothing short of a frequency analysis and possibly a neural net solution would fix this. I think the best bet is human segment voting (which is equipped with frequency analysis and has a neural network).
Sounds may affect people differently. "Loud Noises" is something thats detectable but annoying sounds is opinionated.
If your referring to a live stream then that would need to be done by AI in "realtime".
I think AI would be able to detect and prevent "generic annoying sounds" if the right person was interested in the issue. AI noise cancelling is a thing for microphones maybe it could also work the other way.
amplified and clipped sounds, distortions, single frequency buzzes or beeps or sudden variations or known sounds to provoke reactions
I'd really like to see this for similar reasons. I think other notable use-cases are screamers or even Twitch alerts with VODs, as a lot of streamers make them very excessively noisy.
I think this will be a difficult one to automate without either impeding on the experience of normal viewing, or just not doing a good enough job at times it really matters.
This may also help with things like tinnitus as sudden or unexpected loud noises can often end up triggering it.
Example screamer here:
- Ghost Car (Screamer)
For the noise concerns, I've been looking into audio manipulation techniques to try to remove the annoying beeps while keeping the rest of the audio in #overstimulation channel on discord/matrix
I thunk the best way here is just to set a max audio cap. It would solve the issue
I thunk the best way here is just to set a max audio cap. It would solve the issue
Read the comments above
I thunk the best way here is just to set a max audio cap. It would solve the issue
Read the comments above
Said feature only removes annoying audio. It does not block loud audio.
A max audio volume cap would fix the loud audio, also its quite easy to implement.
You cited the volume of the audio being problematic (breakimg headphones, damaging to ears). An audio cap would solve this