Tape hiss and the clipboard

GoldWave general discussions and community help
Post Reply
Posts: 12
Joined: Thu Mar 27, 2008 7:45 pm

Tape hiss and the clipboard

Post by joedeshon » Sun Oct 17, 2010 11:24 pm

I am converting a pile of cassette tapes to MP3 files. The tapes have a fair amount of hiss on them, as well as some motor-boat fluttering noise. I have been able to reduce most of the noise by copying a portion of the leader tape to the clipboard and then applying noise reduction to the rest of the tape using that sample.

My question is this: I have 10-15 seconds of fairly constant noise at the beginning of each tape. What is a reasonable amount of noise to copy to the clipboard? Is there an advantage to copying a large amount -- several seconds -- so the noise reduction algorithm has a lot of material to work with? Or should I copy only a fraction of a second -- just enough to let it know what noise I need to eliminate?

Is there a standard? Or does it really make any difference? Any thoughts?

Posts: 1164
Joined: Fri Mar 11, 2005 11:15 pm
Location: Washington DC Metro Area

Re: Tape hiss and the clipboard

Post by DewDude420 » Mon Oct 18, 2010 12:02 am

This is a complicated deal...let's see if I can help.

I'm not sure exactly HOW Goldwave creates the NR profile from the clipboard....the process will only vary slightly.

When audio is put in the clipboard and run through noise reduction...goldwave can either a) average out the spectral response of the audio or b) basically keep tabs and record only the peak levels of noise across the spectrum.

Either way..what happens is a profile of the spectrum is created using either of the two methods (honestly, I'm not sure just how Goldwave does it, never bothered to ask Chris), and when the spectral response falls below the threshold it "removes" that chunk of frequency.

Now, in theory...you only need enough audio to get a semi-accurate profile, less than a second usually. However, the more audio you add, the more accurate the profile.

The main thing to rememer is that you want ONLY noise. Surious sounds like clicks...and even the fluttering noise..will cause your profile to have a response that falls outside of the noise range. Clicks will cause a spike in spectral response all across the board while the fluttering noise, which if it contains any low-frequency information, will cause the bass to be reduced.

If the fluttering is actually just pure noise...then that's all going to get averaged out. So start off with small samples and see what the result is like and then increase the amount of noise if needed.

But remember this...regardless of what your FFT settings or anything else...you CANNOT remove ALL of the noise without serious consequences to the rest of the audio, so try to go for reduction rather than elimation.

Post Reply