Page 1 of 1

Removing noise form 78s

PostPosted: Mon Jun 15, 2009 8:44 pm
by mannie.gross
Some time ago I think there was a discussion of removing noise from mono 78 recordings by subtracting left & right channels (which leaves the noise signal) and then subtracting this noise signal from the left & right channels. Was this technique ever implemented in Wave Corrector? Was this technique tried by anyone and, if so, with what results?

Thanks and regards,
Mannie

PostPosted: Wed Jun 17, 2009 9:36 am
by renowden
I'm sure Derek could correct me with the practicalities, but mathematically this is not possible.

L = S + Nl
R = S + Nr

so L - R = Nl - Nr

Subtracting this from the original gives

L' = (S + Nl) - (Nl - Nr) = S + Nr

Which is no nearer S than it was before, all you have done is switched the channels.

PostPosted: Thu Jun 18, 2009 9:22 am
by Derek
Yes, as Rick says, you cannot just subtract the left from right channels to get the noise.

However, you can use the (L-R) difference signal to help detect clicks. This is because the difference channel has no wanted signal (it is cancelled out). This makes it easier to detect the discreet clicks that sit above the general background mush. There is no music in the difference channel to confuse the click detector and cause 'false positives'.

To active this mode, select Auto-Scan Options and click the Advanced button. Set the Click Detect Mode to 'Differential'. Note, although this gives a theoretical advantage, in practice it doesn't give much improvement.

Another option in Wave Corrector allows you to modify the action of the hiss filter when processing 78's. 78's have a lot of hiss-like surface noise but because of the different stylus velocity between the start and the end of the record, the hiss reduces in pitch as the record plays. To account for this, the program offers the option to have separate noise profiles for the start and the end of the record and it interpolates between the profiles when applying the hiss filter. There's a section in the Help file called 'Processing Shellac Records' that discusses this.

regards