Vinyl declicking at Hi-Rez

[Key]

For the record... The correct title is "Judy In Disguise (with Glasses)" by John Fred and His Playboy Band.

Yeah that's the one. I'm just going from memory I haven't listened to that in years to tell the truth.

 

[Bonzodog]

A good representation of spectral techniques in noise reduction is Cedar Retouch.

http://www.cedaraudio.com/downloads/cambridge_handouts/retouch.pdf

Every sound or noise has a spectral "fingerprint" and a record click is very different from a tape splice.

Using these techniques, one can even remove the squeak of a piano pedal without harming the music,
using "photo-shop" like tools to paint out the offending noise.

I was at one post-production house here in Hollywood where they played me a track from a boom
microphone. The track had a door slam along with the actors dialog.

First, they removed the door slam from the dialog, and placed it center channel where the actor was visually.
Then they took the same track, but removed the actors voice from the door slam,
and placed that screen right where the door was! Two separate tracks of audio from a mono source.

 

[Old Quad Guy]

A good representation of spectral techniques in noise reduction is Cedar Retouch.

http://www.cedaraudio.com/downloads/cambridge_handouts/retouch.pdf

Every sound or noise has a spectral "fingerprint" and a record click is very different from a tape splice.

Using these techniques, one can even remove the squeak of a piano pedal without harming the music,
using "photo-shop" like tools to paint out the offending noise.

I was at one post-production house here in Hollywood where they played me a track from a boom
microphone. The track had a door slam along with the actors dialog.

First, they removed the door slam from the dialog, and placed it center channel where the actor was visually.
Then they took the same track, but removed the actors voice from the door slam,
and placed that screen right where the door was! Two separate tracks of audio from a mono source.

That’s awesome. I’ve wondered if the separate musical elements of a Mono or Stereo song could be extracted and remixed as discrete multi-channel music. It sounds like they’re already doing this to some degree. Although if it could be done, would a mix from extracted Mono / Stereo audio truly be considered a legitimate multi-channel mix or something else entirely. I suppose it depends what the end results were. Has this been tried yet?

 

[Key]

That’s awesome. I’ve wondered if the separate musical elements of a Mono or Stereo song could be extracted and remixed as discrete multi-channel music. It sounds like they’re already doing this to some degree. Although if it could be done, would a mix from extracted Mono / Stereo audio truly be considered a legitimate multi-channel mix or something else entirely. I suppose it depends what the end results were. Has this been tried yet?

I've tried a bit of this before. It seemed like a logical step for me since I like to remix and extract samples. You basically take a stereo track and run it through the gamet. Rip it to shreds everywhich way you know how, and then piece it back together in as logical/intuitive a way as you can think of or find through trial and error. It can actually resemble a matrix/decoder to an extent, but isn't limited to real time or linear decoding.

You can use strange logic tricks that most people wouldn't think of like triggering a phase cancelation with a sidechained gate. The effect can be triggered by a super isolated lo fi sound that you don't even hear but can help the other hi fi sounds isolate better etc..

It's like a grey area between a logic decoder and remixing imo. I'm not saying one yields better results than the other - I bet it would be hard to beat some tate/lafayette or other matrix decodes if it was the right mix. But I like the flexibility of the idea.

 

[Bonzodog]

I've tried a bit of this before.

I think we are talking about different things here.

One way the spectral process works is to take an audio signal, and through FFT (Fast Fourier Transform methods of Convolution and Deconvolution), represent the audio signal as a 'picture' and then use 'Photoshop' techniques to 'paint out' the sounds one wants to remove. This “picture” is then re-converted to an audio signal.

To my knowledge, the earliest versions of these techniques were first used by Dr. Thomas Stockham (often called the father of digital recording, google him) in his “Blind Deconvolution through DSP” methods of removing the orchestra from Enrique Caruso recordings. He then recorded new stereo orchestral scores for a 1976 RCA vinyl release of these early wax cylinder recordings.

More recently, I understand that advancements of these techniques have been used to isolate the vocals for new “Duet” recordings from dead artists such as Frank Sinatra and Nat King Cole.

It sounds to me like you are using analog “mash-up techniques” rather than DSP.

 

[Key]

I think we are talking about different things here....
It sounds to me like you are using analog “mash-up techniques” rather than DSP.

I wasn't actually talking about the specific technique you brought up but more the idea of isolating sounds in general. Although I worked at a photolab and did photoshop for a while and had wondered if something similar to what you are talking about existed or not. And no I'm not talking about analog but digital.

 

[See Why Audio]

Hi all,

Now that I've got Audition 3.0, I find one of my most-used presets no longer works.

In AA 1.5 I was used to being able to marquee certain frequencies of a click and, using a favourites preset, hit 'fix single click now'. I would use this technique to first fix the high frequencies with a narrow marquee and then move to the lower frequencies and fix them with a wider marquee, thus reducing damage to the surrounding higher frequencies while fixing the 'overhang' of a big click. Now I just can't do it! Grrr.

There is a problem with using this technique in AA 3.0:
When you do the same procedure, the resulting audio contains a 'ringing' artefact above or below the marqueed area you were fixing. So: If you marquee the high frequencies of a click and 'fix single click' you get a ringing artefact just below the bottom edge of the marquee, it's the same for low frequency marquees, you get a ringing artefact above the marquee!

Check the image.

What a right royal pain. Obviously the new AA 3.0 treats the data differently in this respect.
I have found however that isolating clicks in standard waveform view now gets treated with less aggression, or it seems, more cleverly, and it is usually possible to 'fix single click' in wave view and not get a damaged section. Problem is I now have to view-toggle waveform/spectral twice to get the job done. Humph. More RSI.

Anyone else found this? Any solutions?

Colin