Premier Rock Forum

engineering
fixing

i feel ambivalent

i recognize potential but i feel a bit...

icky

jordanosaur wrote:I agree with the above statement -

Ethical usage aside, it's pretty amazing what they have been able to do from an engineering standpoint. That being said, I think it's probably going to be abused in the most disgusting ways.

Dontfear:

Out of curiosity, what would you use this for specifically? I'm not railing you, just curious.

I work primarily in the B room at the studio where I work. This is the fast, cheap and dirty room. We don't end up loaning our clients perfectly intonated guitars to use, so every now and then, there's a chord that jumps out at me as having a badly intonated note, but it's not like I can stop this punk band and say "Wow, you guys... I am going to need you to stop and take that guitar to a luthier and find that bad bit of intonation there..."

Being able to fix this on the fly will make my life a lot easier.

that's.... crazy.

I don't know, the reactionary side of me thinks "wrong wrong wrong", but then as someone who's ruined great takes of ten minute long songs by hitting one bum note, I can see the appeal....

I have a lot of doubts about the usefulness of this. Autotune already sounds pretty bad, and you have total isolation of a monophonic vocal track. How does it actually know what harmonics are associated with a root frequency? It can't know- it can only make intelligent guesses. There will always be residual poo poo left over. I don't really think this is a revolutionary idea at all. It is just a natural extrapolation of existing technologies. I think it will be cool as a special effect but I think it will degrade the sound too much. Musical instruments are far too complex, they don't produce the same exact harmonic series for different notes. Unless you want your guitar to sound like an old general MIDI keyboard...

I would have no use for any of the applications discussed in the (excruciating) film. I do however spend a fair amount of time decomposing sounds with filters and such, so any new way to reach in to a complex sound and grab a part of it strikes me as potentially interesting. I wonder how it works, and how well.

FFT monster it is I bet too, that's how it's able to sound so realistic vs. something like autotune.

I have a feeling it doesn't handle distortion very good at all though, if at all. None of the samples they showed had distortion either.

projectMalamute wrote: I wonder how it works, and how well.

My best guess is that it uses FFT to analyze the spectrum and determine what frequencies possibly make up the chord. It would most likely recreate the chord with simple sine waves and then use a similar spectral analysis to scale and impose the harmonic content (vocorder-like) back on to each voice of the chord of sine waves it created.

Where the fun would be is in determining where the break point is when trying to discern the voice components of the chord from the harmonic content that makes up the timbre. This could be a humoungus gimmick box if it is used to change distortion/harmonic content into voices.

Any other ideas?

Hi,
Great tool.

jordanosaur wrote:Ethical usage aside

Common' this is music ... why talk about ethic ?
/Nick.

ktone wrote:

projectMalamute wrote: I wonder how it works, and how well.

My best guess is that it uses FFT to analyze the spectrum and determine what frequencies possibly make up the chord. It would most likely recreate the chord with simple sine waves and then use a similar spectral analysis to scale and impose the harmonic content (vocorder-like) back on to each voice of the chord of sine waves it created.

Where the fun would be is in determining where the break point is when trying to discern the voice components of the chord from the harmonic content that makes up the timbre. This could be a humoungus gimmick box if it is used to change distortion/harmonic content into voices.

Any other ideas?

The vocodor like re-synthesis angle is an interesting thought. First thing that popped in to my head was the use of a series of fairly narrow band-pass filters. Problem with that is you would get lots of stuff from other voices, think about how much overlap you would get just playing an octave or perfect 5th. I was thinking in terms of filtering stuff out, but you are probably right that the process is more one of analysis and then recreating the stuff from scratch.

I'm just afraid of what's going to happen on 8/29 when it becomes self aware