unforum

[Euchre]

Water?

[Tracker_56]

[FilmEdge]

Wet just makes it better!

[vapor]

[jiblington]

@vapor

where did you get the information on convolution from? everything ive been taught thats just natural reverb emulation. sounds like a good read for a techy like me though.

[vapor]

[SeekerX]

Lol Ras, should have known that a Phi Gesture couldn't look anything but beautiful .
Nice find Vapor. Never heard of convolution before, probably that came after my time. Not sure if you could apply it to the sound for any good effort. From what I understand you, like with de-vocoding, ought to know the decryption-key. Like with the de-vocoder the carrier -signal, with convolusion you'd need to know the signal with which the actual soundfile has been processed to get the "true" result.

[vapor]

[SeekerX]

Hmmm. Am I not right in my understanding that Convolution is kind of overlaying/processing the acoustics of one "space" onto a sample or whole piece of recording?

Quote from the Wiki:

[vapor]

I don't completely understand the math behind deconvolution, but the short answer to your question is no, you don't need the original impulse tone to be able to extract it. Only the final output and a test tone are needed. I used a sine wave sweep as my test tone but you could also use a burst of silence.

[SeekerX]

Actually I haven't read anything much of your previous links. Probably I should have but my understanding tells me that a software needs to know what it should look for. OK, maybe it can find patterns and from pre-programmed formats/algorithms it might be able to more or less find the right stuff to filter out. But until you yourself can't definitely tell it what to filter out, it's just guessing.
Convolution is the art of reprocessing a natural reverb onto a digital file by software, right? A natural reverb consists of a lot of factors. "First responders", "damp", "depth", threshold", etc. etc. etc.. An original and exact copy of a given natural reverb is probably the most difficult task in audio engineering, if you want to drive it to extremes. Not even good men are able to do exact reproductions (well, they weren't 10-15 years ago). So, I doubt an automated software could do better and do even the opposite (re-calculating without the actual source).

[GeniferB]

This may be nothing, but I just had a thought.

In regards to the stuff that was found yesterday, there was a zip code that led us to West Virginia, near the same area as the Mothman sightings. Has anybody compared our sounds from the Braegen14 file with the sounds that a moth would make? It might be a long shot, but it just occurred to me.

I am at work and don't have sound on my computer so I can't look into it now.

[vapor]

@SeekerX, believe me I was skeptical too. And apparently I'm not the only one. One of the websites I came across while researching was a Math blog in which mathematicians put deconvolution software to the test. They tested it by convolving their of sound file, then taking that file into their test software and deconvolving it. Sure enough, they got their impulse file back. They even took it a step further and used the impulse file from their test to re-convolve and compared their 2 files. There was minimal degradation.

I'll have to see if I can find that blog. It's more of an academic afterthought at this point, though. I don't think convolution is relevant.

[Froid]

[jiblington]

i havent had time yet too read the pages on convolution up above(but i defo will), but i was told in a very long lesson about artificcal reverbs that the beatles were recording reverbs and using them on things they had recorded using analogue techniques back in the 70s, not sure how they did this exactley but if the method differs radically from how it is done digitally(which i presume it should) then you may get different results from doing it digitally, not that i think jj would expect people to have some old reel to reel tape decks and a decent analogue workstation to undo such tinkering but its interesting, well to me at least and fits the convo.