5 ms·

Id dream of a raspberry pie plugin that has a mic & IR transmitter that just listens to ads or any semblance of it and just hits mute the moment the confidence

Id dream of a raspberry pie plugin that has a mic & IR transmitter that just listens to ads or any semblance of it and just hits mute the moment the confidence interval goes beyond 90% based on the rise in ampltitude. False positives would be straining though, and finding the moment to un-mute is also a bit difficult perhaps. But given that all ads have their loudness dialed up to 11 in relation to the actual program, it might be easy enough to just listen to a large drop in amplitude across the frequency spectrum to hit un-mute. This can be completely off-the-grid with incidental updates if so desired. Could even put some federated learning into the bag for a slight dystopian flavour - and that sweet VC allure - to also listen to people hitting mute & unmute as a semi-reliable labelling feature. Just sample the time window recording of the frequency spectrum and it may be good enough already. It wouldn't be foolproof but just good enough I guess. I'd be willing to pay 200$ for a ad silencing IR transmitter with >90% recall.

jaycar 5y ago
you should fork sponsorblock.
stonogo 5y ago
Have at it: https://circuitcellar.com/research-design-hub/projects/build-a-tv-commercial-killer/ https://circuitcellar.com/research-design-hub/projects/build... The device you would pay $200 for used to cost $40: https://bgr.com/general/mute-tv-commercials-automatically-how-to/ https://bgr.com/general/mute-tv-commercials-automatically-ho...
squarefoot 5y ago
AI could come to help: a powerful enough dedicated hardware could decode HDMI signals, match the audio/video with the known fingerprint of an ad, substitute it with say a tropical beach or some mountains with a soft jazz music background and a small countdown in a corner "ads end in: mm:ss", then resume the program at the predicted end. The fingerprint would be taken by the users by pushing a button to mark each spot start and end so that the AI can recognize it and upload to a central server that would aggregate all data. The system should have a buffer long enough to allow watchers to rewind to the exact start and end of the spot, and to allow the AI to analyze the data before showing it to the watchers, however a very long buffer could be even more handy to wait for the entire program to end, then process it so that every ad is removed and the program joined back together with no interruptions. Doable? Yes, but would need a lot of work and some money for the hardware. Probably also overkill since pretty much all programs are already available ads free on decentralized torrent swarms.
- badestrand 5y ago
  You don't need AI for that. 20 years ago my neighbors had a small device that would automatically mute the TV whenever the ads started playing and unmuted when they finished. No idea how it worked but for sure it didn't have a fancy neural network inside.
  - squarefoot 5y ago
    That device sensed the higher average sound levels at which ads are usually broadcasted, and it's also easy to build. The problem is those ads that have or start with lower sound levels which would be impossible to spot without matching them with a known audio/video pattern.
  - ComradePhil 5y ago
    It looks in the video signal for the TV channel logo. No logo means ads.
- vanilla_nut 5y ago
  Sinilarly, this is the only “AR” I ever want in my life: glasses that detect billboards/subway/bus ads and either “magic erase” them or just replace them with images of my choice.
withinboredom 5y ago
MythTV (open source DVR) would skip commercials in the recording. So I’d just watch my shows an hour later without commercials.
eru 5y ago
I think the ads don't really have the absolute loudness dialed up, but use a lot of dynamic compression instead. That's what (used to?) happen to music during the so-called loudness war. See https://en.wikipedia.org/wiki/Loudness_war https://en.wikipedia.org/wiki/Loudness_war
williamtrask 5y ago
This is a really nice federated learning usecase. I work with an open source FL community. Will add it to the list.
horsestaple 5y ago
Would be pretty easy to do in this eastern EU country. We have laws that require stations to separate ads from content, so there is always a short animation announcing the start and end of ads.
rcarmo 5y ago
I think Douglas Adams wrote about something very much like that, but I can't place it. Maybe it was Philip K. Dick?
- gilmore606 5y ago
  You are thinking of Carl Sagan's novel Contact, in which the rich benefactor HR Hadden makes his fortune on a device to mute TV ads.
denton-scratch 5y ago
> based on the rise in ampltitude The apparent rise in amplitude is horribly annoying; but I understand that actually increasing the volume during ad-breaks is forbidden (at least in the UK). Instead, they tailor the audio spectrum to create psycho-acoustically the impression of increased loudness. Automated ad-filtering on broadcast TV seems to be a hard problem. Some filters look for the channel logo screen at the start and end of the ad-break; but (a) you have to set up each channel; (b) "sponsors" now book slots at the start and end of the break, before and after the channel-logo screen. The best solution I've found is to record anything I want to watch in advance; then to skip forward 5 minutes when the break starts, and FF/RW if I land in the wrong place. I'm a bit surprised that so many people seem to be buying big-screen TVs, but apparently not STBs with hard-disk recorders.
- plasticchris 5y ago
  I used to have a DVR that would auto skip commercials on recorded shows but that was probably using metadata. It was annoying that it didn’t do the same live.
lazyeye 5y ago
Id also want a visual "mute" where the picture shrinks to tiny square in a corner of the screen whilst the main screen is either blank or shows something pleasant and relaxing
adrianmonk 5y ago
> federated learning One approach that occurred to me is to take advantage of the fact that same ads are played in different TV shows. Suppose you record hundreds or thousands of hours of TV video/audio, then compare it to your program guide data. In theory, you should be able to detect that some segments of the video/audio correlate very strongly with what show you're watching (occurring only there), whereas other segments of video/audio appear all over the place in different programs. And use fingerprinting to avoid having to store/transmit all that video/audio data. This is not that different from what Shazam does with music recognition, although their job is easier because their audio data is naturally segmented into songs, whereas for ad detection you have to find segments. But I bet there are algorithms to efficiently find common sub-sequences (maybe from DNA bioinformatics?). Once you have fingerprints of commercials, you distribute a database of these to viewers, and their device uses it to detect commercials while watching. --- For programs that air more than once, you could invert this and look for segments of video/audio that are common between the airings. The rest is more likely to be commercials, especially if you've got a movie that airs on different networks. Or if you can see a show/movie through on-demand, maybe you'll get different commercials on repeated viewings. --- Another approach is facial recognition. If you see Flo from Progressive, it's probably an ad. You can maybe even recognize new, unknown ads this way. If you want to go really crazy, mine IMDB for TV/movie titles, the cast that appear in each, and their photos, then recognize non-ad video footage based on who you know is supposed to be in that episode.

jaycar 5y ago

you should fork sponsorblock.

stonogo 5y ago

Have at it: https://circuitcellar.com/research-design-hub/projects/build-a-tv-commercial-killer/ https://circuitcellar.com/research-design-hub/projects/build... The device you would pay $200 for used to cost $40: https://bgr.com/general/mute-tv-commercials-automatically-how-to/ https://bgr.com/general/mute-tv-commercials-automatically-ho...

squarefoot 5y ago

AI could come to help: a powerful enough dedicated hardware could decode HDMI signals, match the audio/video with the known fingerprint of an ad, substitute it with say a tropical beach or some mountains with a soft jazz music background and a small countdown in a corner "ads end in: mm:ss", then resume the program at the predicted end. The fingerprint would be taken by the users by pushing a button to mark each spot start and end so that the AI can recognize it and upload to a central server that would aggregate all data. The system should have a buffer long enough to allow watchers to rewind to the exact start and end of the spot, and to allow the AI to analyze the data before showing it to the watchers, however a very long buffer could be even more handy to wait for the entire program to end, then process it so that every ad is removed and the program joined back together with no interruptions. Doable? Yes, but would need a lot of work and some money for the hardware. Probably also overkill since pretty much all programs are already available ads free on decentralized torrent swarms.

badestrand 5y ago
You don't need AI for that. 20 years ago my neighbors had a small device that would automatically mute the TV whenever the ads started playing and unmuted when they finished. No idea how it worked but for sure it didn't have a fancy neural network inside.
- squarefoot 5y ago
  That device sensed the higher average sound levels at which ads are usually broadcasted, and it's also easy to build. The problem is those ads that have or start with lower sound levels which would be impossible to spot without matching them with a known audio/video pattern.
- ComradePhil 5y ago
  It looks in the video signal for the TV channel logo. No logo means ads.
vanilla_nut 5y ago
Sinilarly, this is the only “AR” I ever want in my life: glasses that detect billboards/subway/bus ads and either “magic erase” them or just replace them with images of my choice.

withinboredom 5y ago

MythTV (open source DVR) would skip commercials in the recording. So I’d just watch my shows an hour later without commercials.

eru 5y ago

I think the ads don't really have the absolute loudness dialed up, but use a lot of dynamic compression instead. That's what (used to?) happen to music during the so-called loudness war. See https://en.wikipedia.org/wiki/Loudness_war https://en.wikipedia.org/wiki/Loudness_war

williamtrask 5y ago

This is a really nice federated learning usecase. I work with an open source FL community. Will add it to the list.

horsestaple 5y ago

Would be pretty easy to do in this eastern EU country. We have laws that require stations to separate ads from content, so there is always a short animation announcing the start and end of ads.

rcarmo 5y ago

I think Douglas Adams wrote about something very much like that, but I can't place it. Maybe it was Philip K. Dick?

gilmore606 5y ago
You are thinking of Carl Sagan's novel Contact, in which the rich benefactor HR Hadden makes his fortune on a device to mute TV ads.

denton-scratch 5y ago

> based on the rise in ampltitude The apparent rise in amplitude is horribly annoying; but I understand that actually increasing the volume during ad-breaks is forbidden (at least in the UK). Instead, they tailor the audio spectrum to create psycho-acoustically the impression of increased loudness. Automated ad-filtering on broadcast TV seems to be a hard problem. Some filters look for the channel logo screen at the start and end of the ad-break; but (a) you have to set up each channel; (b) "sponsors" now book slots at the start and end of the break, before and after the channel-logo screen. The best solution I've found is to record anything I want to watch in advance; then to skip forward 5 minutes when the break starts, and FF/RW if I land in the wrong place. I'm a bit surprised that so many people seem to be buying big-screen TVs, but apparently not STBs with hard-disk recorders.

plasticchris 5y ago
I used to have a DVR that would auto skip commercials on recorded shows but that was probably using metadata. It was annoying that it didn’t do the same live.

lazyeye 5y ago

Id also want a visual "mute" where the picture shrinks to tiny square in a corner of the screen whilst the main screen is either blank or shows something pleasant and relaxing

adrianmonk 5y ago

> federated learning One approach that occurred to me is to take advantage of the fact that same ads are played in different TV shows. Suppose you record hundreds or thousands of hours of TV video/audio, then compare it to your program guide data. In theory, you should be able to detect that some segments of the video/audio correlate very strongly with what show you're watching (occurring only there), whereas other segments of video/audio appear all over the place in different programs. And use fingerprinting to avoid having to store/transmit all that video/audio data. This is not that different from what Shazam does with music recognition, although their job is easier because their audio data is naturally segmented into songs, whereas for ad detection you have to find segments. But I bet there are algorithms to efficiently find common sub-sequences (maybe from DNA bioinformatics?). Once you have fingerprints of commercials, you distribute a database of these to viewers, and their device uses it to detect commercials while watching. --- For programs that air more than once, you could invert this and look for segments of video/audio that are common between the airings. The rest is more likely to be commercials, especially if you've got a movie that airs on different networks. Or if you can see a show/movie through on-demand, maybe you'll get different commercials on repeated viewings. --- Another approach is facial recognition. If you see Flo from Progressive, it's probably an ad. You can maybe even recognize new, unknown ads this way. If you want to go really crazy, mine IMDB for TV/movie titles, the cast that appear in each, and their photos, then recognize non-ad video footage based on who you know is supposed to be in that episode.