High pitch tones - removal of

A recording via TASCAM DP008 digital recorder has picked up (or generated) a pair of high-pitch tones over spoken voice. Around 4k-5kHz, and about 0.3secs duration. They produce very pronounced clipping.

Compression helps a bit

When I try to remove the tones with Audacity (v2.0.3) LPF plugin it seems like the problem replicates. The pair become 4 or 6 and they are more broadly spaced, timewise.
I can’t manage to drop the notch filter on target either. (Toshiba laptop is the computer in use with Win7)

Any suggestions?

It may help if you can attach a short audio clip.

Around 4k-5kHz…

…When I try to remove the tones with Audacity (v2.0.3) LPF plugin it seems like the problem replicates. The pair become 4 or 6 and they are more broadly spaced, timewise.

I assume you are seeing the tones on the spectrum (Analyze → Plot Spectrum)?

They produce very pronounced clipping.

Clipping will generate harmonics, but these will be at higher frequencies, so a low-pass filter should reduce these too.

Sample now attached.

Yes - the clipping shows up as three broad red bars which I can get rid of almost completely via compression.

(I don’t have the option of re-recording. Note to self - listen to the whole thing carefully before saying goodbye to the speaker!!)

Progress today…

No joy with effects - normalise …compression…level… reshapes the audio but still leaves tones, and LP filter seems to make them worse.

However…changing sample rate from 44100 to 48000 seems to do the trick. Still leaves a visible and audible rough spot but some fade-sandpaper may sort that (or BBC can sort it out before the broadcast!).

The original WAV file from the recorder is fine when played on the standard Windows media player. It corrupts when played on Audacity, but only at this one place in the full 3 min recording.

Sorry, missed a query.

The freq analysis does not show a nice clear spike or octave sequence, even on the sample area where the tone is audible.

I’m at work right now, so I can’t listen to the file or do much experimentation. I’ll try some stuff when I get home later…

Yeah, looking at the spectrum it looks like broadband noise and nothing jumps-out at 4-5kHz. Looking at the waveform, it also looks like there is a big low frequency (or subsonic) component.

We may be able to make some improvement, but I don’t think you can completely fix it.

The original WAV file from the recorder is fine when played on the standard Windows media player. It corrupts when played on Audacity, but only at this one place in the full 3 min recording.

Weird! You may be able to play the file in Windows Media Player while re-recording in Audacity (or with another recording program). [u]This page[/u] gives you several ways of recording what you are hearing from your computer speakers. (Some soundcard drivers don’t support recording the speaker-output directly, but there are software & hardware alternatives.)

However…changing sample rate from 44100 to 48000 seems to do the trick.

Great! But, I don’t see why that would make a difference. Does that mean everything is OK now? Did you just change the Project Rate in the lower left hand corner of the Audacity window to 48,000 and re-export?

In the final run all I really did was change the sample rate to 48000 from 41400. Tone disappeared. Another post had said something about this, but not for exactly my problem.

This was the first time I’ve recorded spoken voice using the recorder’s inbuilt mikes - usually the recorder is for choral concerts and I have a couple of half-decent condensor mikes. Otherwise the process was identical to normal - download the event in total and take WAV track(s) straight into Audacity for processing. I can’t think of any physical/external reason for the disturbance observed. Since the sample rate change is at least implicated, I’m assuming some quirk in software or soundcard.

Thanks for your time/guidance. At least I know I’m not doing or not doing something daft.