continuous tone removal problem

[steve]

This is a section of the Plot Spectrum window - I've turned off the grid to make it easier to see.
This is an analysis of the silence at around 3.5 to 4.5 seconds.
The thing to notice is that thin spike, which is the whistle.

Screenshot.png (3.25 KiB) Viewed 351 times

Placing the cursor exactly on the spike will give a fairly accurate readout of the frequency, particularly if you set the "size" setting high.
A rough and ready notch filter setting: Frequency = 1775 Hz and a width (Q) = 8 will remove the whistle very effectively.

[Spamlet]

That's brilliant Steve, thanks for the excellent advice yet again.

I've rerecorded it in mono to make things easier (there wasn't much in the way of stereo effects in it anyway).

I noticed an extra buzz second time around, and after some head scratching found the tape head was picking up interference from a lighting dimmer switch. Something other readers might like to bear in mind.

Cheers,

S

[DickN]

Another trick for tuning a notch filter is to use the 'single band parametric' effect (an equalizer). Once you've found the approximate frequency with Analyze->plot spectrum, you can tune the filter by previewing repeatedly with 'single band parametric' set to boost the frequency to which you're setting it. Keep reducing the bandwidth as you home in on the exact frequency. With width set to 0.002 octaves, you can attenuate a single frequency quite a lot without doing much collateral damage to voice. I do this routinely to a squeaky paddle fan in our church services.

[Spamlet]

Cheers Dick,

What a lively and informed forum this is!
As I have hundreds of tapes to deal with I may be back!

S