add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
browse anonymously,
or get an account
and write.
register,
|
|
|
Please log in.
Before you can vote, you need to register.
Please log in or create an account.
|
De-crackle
internet service to remove noise from old recordings | |
A website that allows owners of old and damaged vinyl records to upload high quality audio files. When several recordings of the same piece of music have been uploaded a cleaned up version is made by comparing the recordings and removing the parts that dont match up. If the copyright needs to be protected
the new cleaned up version could be emailed only to people who upload the track.
[link]
|
|
Each person would need an archival quality record player. Also, since there is nothing to compare the recordings against besides each other, the server would basically have to make an educated guess as to which samples are the best. |
|
|
For most recordings that are common enough to get a sufficient presence on the server, there already exist remastered versions. |
|
|
Anything that's more rare than that really should be recovered professionally. There's a lot of whizzbang technology that the pros have that can extract the best possible data from a single record even if damaged. |
|
|
I like the theory here though, however I think it would be better applied in something else. Bun anyway. |
|
|
Remastered recordings sound horrible. I try not to buy them, especially for classical music. This idea OTOH would probably work well. |
|
|
Oh and just to be pedantic (since that's popular around here) you wouldn't want to use MP3 because its lossy and audiophiles don't like it. |
|
|
In the same way that you can take half-decent astronomical photos (apparently) with a web-cam and "stacking" software, perhaps it may become possible to make a hifi recording of an event (gig, political speech, gnome infestation) with multiple low-res microphones arranged variously, and perform some kind of signal-recovery operation on them too. |
|
|
Could the same thing be done in (perhaps interplanetary) broadcasting to ensure that signal-to-noise ratios remain acceptable? |
|
|
[zen_tom], Id think low-res microphones would not result in high-quality sound, regardless of the number of gnomes. |
|
|
//mp3 recordings//
As [Spacecoyote] mentioned, you need to start with a lossless signal. So when the idea says "mp3", I'm assuming it's just a generic term for "audio file", maybe shorthand for "PCM". Otherwise, processing will result in audio that sounds worse than before, albeit with less "noise". |
|
|
Do people who rip high-quality files really need a special web site to "clean up" their audio? If the noise is unacceptable, compression artifacts are, too. Otherwise, this idea is like buying a wide-screen TV for the "flawless picture", and not noticing that the aspect ratio is whacked. |
|
|
Even though it's web based, there doesn't have to be compression artifacts. To save bandwidth the audio could be compressed with a lossless codec such as shorten or FLAC. The compression ratio is actually pretty good, at least compared to uncompressed. |
|
|
Since running this in a traditional webserver way would cost loads of $$, I suggest a peer-to-peer approach, perhaps using the Bittorrent protocol. |
|
|
youre right spacecoyote using mp3s would be a bit pointless. I dont really know about audio files for audiophiles or bit torrent so i'll leave the details out. Thinking about what zen_tom said, maybe if multiple microphones were positioned at the same distance from a sound source other sounds in a room would reach each microphone at different times and wouldn't match up. if the original idea would work it could be applied in this case to remove unwanted noises and sound reflections from a live recording. |
|
|
You could sell the removed crackle from valuable recordings to new artists who want that antique sound. |
|
|
There are several systems that will do this better, from some cheap CoolEditPro plugins (actually, surprisingly good), though Powercore Restoration Suite (excellent), right up to the full-blown CEDAR System which is slightly better than anything else and therefore ten times as expensive. The main difference between what you suggest and these systems is that they all require an operator who makes decisions on how to use the tools while listening to the output. |
|
|
This is in essence quite a clever idea, but it needs to involve a pair of ears somewhere. Combining several different music files involves many variables - which tempo is closest to what it "should" be, how much of each track to use, eq, compression, expansion, the list goes on. Computers are just not qualified to make these decisions as they have no concept of what sounds good. |
|
|
Wasn't that how UNIVAC predicted Dwight Eisehower's victory? |
|
|
This is a very good idea and exactly what is needed to
recreate original-quality copies of 78's. |
|
|
It would work best on recordings of disks from the same
master that should be identical, then once you have
enough copies to do a concensus function on all
waveforms, you should be able to extract the common
waveforms minus the noise and scratches from the
various old disks that the recordings came from. |
|
|
Note they have to be recordings of different disks, not
multiple recordings of the same physical object. |
|
|
Definite bun. This is HDR photography for audio -
restoring the original sound, not trying to approximate it
with various filters and guesses as to missing content. |
|
|
(Actually turns out that there is not a direct photography
equivalent of this - see the post on "Old print
restoration") |
|
|
I'ld like to see this implemented. |
|
| |