I had been thinking about something like this, but in the context of recreating riot crime scenes. After the last riot in Toronto the police received over 1M stills and thousands of hours of relatively low quality cell phone video. The main task for them is to connect different shots of individual offenders over time to build a) a coherent story demonstrating premeditation, b) an unbeatable description and connect it to some identifying info that might not necessarily be present at the exact time of the offense.
Show HN: Entire concerts algorithmically "reconstructed" from YouTube videos
61–70 of 132 posts
Re: Show HN: Entire concerts algorithmically "reconstructed" from YouTube videos
#62Earlier quoted context omitted.
That is just so much harder; though I haven't looked at the website yet, seems to be overwhelmed right now. Besides, I would want one consistent audio signal instead of one that varies in noise, volume, whatever. Video from different sources is alright though, since we are used to switching scenes and cameras all the time.
The idea I'm talking about should provide a single consistent audio signal. I know nothing about audio processing, but it seems like it should be possible to take multiple bad audio signals and combine them into one signal that's better than any constituent audio source. Perhaps one audio source captured low frequencies well, while another captured higher frequencies better.
You can certainly automate crossfaded audio between multiple sources to try to get the cleanest copy, but it's hard. For instance, how do you decide whether it's noise or the letter "s" or the "chk" of a pick across muted guitar strings? The heuristics for "better than any constitutent audio source" can be extremely nuanced, algorithmically intensive, and still difficult to pin down, akin to speech recognition. Speaking purely to SaaS'y automated purposes, natch.
Typically what it seems you're talking about for audio here is similar to a matrix mix in the amateur/live audio world. People have been (manually) mixing soundboard audio with audience-recorded audio to improve the audio quality of recorded shows for some years now.
Re: Show HN: Entire concerts algorithmically "reconstructed" from YouTube videos
#63Earlier quoted context omitted.
Courts ruling on fair use are supposed to take into account "the amount and substantiality of the portion used in relation to the copyrighted work as a whole". Automated copyright detection schemes in YouTube may try to do some semblance of fair use by allowing snippets of copyrighted content under 30 seconds.
Honest question - does the copyright still cover a live work when the original recording was permitted in the first place?
Yes it does. Just because the recording was permitted doesn't mean that the recorder has the rights to publish or replay the recording.
Another example: owning a DVD does not mean you can then stream the DVD over Justin.tv.
The thing about rights is that infringement must be claimed by the rightsholder for any action to be taken. Once a rightsholder claims infringement then the content/stream/video must be shut down or removed, and if the site or person continues to infringe, THEN action can be taken.
This is one of the good things about the DMCA and the reason YouTube, Soundcloud and even justin.tv can continue to operate. It is also exactly what the much maligned SOPA bill is trying to change - for the worse.
btw, you should sign a petition against that bill on votizen or similar.
They climbin in yo' windows, snatching your youtubes up.
Re: Show HN: Entire concerts algorithmically "reconstructed" from YouTube videos
#64First thing I thought of was the Nine Inch Nails fan project "This One Is On Us". It works. Very cool and I wish you all the best.
Re: Show HN: Entire concerts algorithmically "reconstructed" from YouTube videos
#65Re: Show HN: Entire concerts algorithmically "reconstructed" from YouTube videos
#66Wow. Many thoughts arose: - Is this more, less, or equivalently legal (copyright-wise) to just posting the youtube videos on youtube in the first place? It seems like the act of compositing them in this way would be significant, but maybe not? - The normalisation of audio could use some work. SwitchCam seems to do something (the youtube volume on each clip was different for me), but it didn't quite work. - I'd be rea…
Courts ruling on fair use are supposed to take into account "the amount and substantiality of the portion used in relation to the copyrighted work as a whole". Automated copyright detection schemes in YouTube may try to do some semblance of fair use by allowing snippets of copyrighted content under 30 seconds.
just a sidenote - if bittorrent software ensure that no peer-to-peer interaction exceeds 30 sec. for any given title, would MPAA/RIAA/etc... still be able to claim "illegal dowload"?
Re: Show HN: Entire concerts algorithmically "reconstructed" from YouTube videos
#67So if you can do this with concerts, presumably you can do it with any collection of video shot in the same place, if there's enough audio in common.
That makes me think that all kinds of crowd video (protests, speeches, etc) should be stitchable, which seems like it would open larger (and potentially morally ambiguous) markets.
Re: Show HN: Entire concerts algorithmically "reconstructed" from YouTube videos
#68Earlier quoted context omitted.
That is just so much harder; though I haven't looked at the website yet, seems to be overwhelmed right now. Besides, I would want one consistent audio signal instead of one that varies in noise, volume, whatever. Video from different sources is alright though, since we are used to switching scenes and cameras all the time.
The idea I'm talking about should provide a single consistent audio signal. I know nothing about audio processing, but it seems like it should be possible to take multiple bad audio signals and combine them into one signal that's better than any constituent audio source. Perhaps one audio source captured low frequencies well, while another captured higher frequencies better.
This seems unlikely to be possible. If you haven't captured any frequencies above 15kHz (which an average cell phone mic is unlikely to do), no amount of averaging, filtering, or combining will get them back. There will also be a considerable amount of distortion, since concerts tend to be so loud that even one's ears are distorting. Good luck separating physical distortion in the mic, limiter distortion in the analog or DSP stage, and clipping distortion at the ADC.
I think the best you could do is use the video to determine where someone was standing, and try to reconstruct some of the stereo information based on multiple recorders.
Re: Show HN: Entire concerts algorithmically "reconstructed" from YouTube videos
#69Cool idea but when I tried "Arcade Fire - Austin City Limits" 2011 http://switchcam.com/event/arcade-fire-austin-city-limits-20... it was far from correct.
Re: Show HN: Entire concerts algorithmically "reconstructed" from YouTube videos
#70Earlier quoted context omitted.
That is just so much harder; though I haven't looked at the website yet, seems to be overwhelmed right now. Besides, I would want one consistent audio signal instead of one that varies in noise, volume, whatever. Video from different sources is alright though, since we are used to switching scenes and cameras all the time.
The idea I'm talking about should provide a single consistent audio signal. I know nothing about audio processing, but it seems like it should be possible to take multiple bad audio signals and combine them into one signal that's better than any constituent audio source. Perhaps one audio source captured low frequencies well, while another captured higher frequencies better.
If both channels have a similar spike at the same frequency at the same time, it is probably part of the signal (not noise), so combine those, and dampen all others. This would cover your case, if the other channel had enough of the low/hi freq of the other to relate them. I reckon Shannon looked at exactly this in developing Information Theory (for telephone signals on flaky lines), and it's probably all textbook stuff now.