Removing pesky ringing sound generated by Blue Snowball mic

[featheredtar]

Hi. What would you recommend for removing this pesky ringing sound that my Blue Snowball microphone generated in this interview? I have experimented with Equalization and Noise Removal. With Equalization I am having trouble removing that specific sound without severely degrading the overall speech quality, and with Noise Removal the sound is removed during periods of silence, but then repeatedly comes back with each new word spoken. If you have any advice as to specific settings to use, it would be much appreciated.
Mac 10.9.2, Audacity 2.0.5.0, from ZIP

High-pitched noise with only ambient noise
https://dl.dropboxusercontent.com/u/112 ... noise1.wav

High-pitched noise with speech
https://dl.dropboxusercontent.com/u/112 ... noise2.wav

[billw58]

Apply:
High Pass Filter, Rolloff 24 dB, Cutoff Frequency 10 Hz (could probably use 20 Hz or even 40 Hz - higher values will remove more low-frequency noise)
Notch Filter, Frequency 60, Q 10
Notch Filter, Frequency 120, Q 10

Now apply Noise Removal:
Select a noise-only portion, then Effect > Noise Removal > Get Noise Profile
Select the entire track, then Effect > Noise Removal
- set Noise Reduction to 12 (a higher value will remove more noise at the expense of "tinkly bells" artifacts), then OK to remove the noise.

How I determined this:
I selected all of the noise-only track and did Analyze > Plot Spectrum. That showed the low-frequency noise, and the two spikes at 60 and 120 Hz. It also showed the "ringing" spikes, but there were too many to use the Notch Filter to remove them all. Getting the low frequency noise and the two low frequency spikes out of the way made it easier for Noise Removal to deal with the ringing.

-- Bill

[kozikowski]

Noise Removal the sound is removed during periods of silence, but then repeatedly comes back with each new word spoken.

Yes. Noise Removal is not a gift from the heavens and in severe cases just changes the problem to a different problem.

The Smoothing setting controls how Noise Removal works around speech. That's the gating effect. The older Noise Removal implementation worked like you think it should by automatically taking the noise out of everything, but aside from a small fan club, most of us found it completely unworkable. Voices would turn to trash very easily and the sliders that controlled it would always run at one end of their range.

[Getting headphones]
Frying mosquitoes. You win actually. On my machines it appears as a tiny hissy warbling scream. It only happens on USB microphones or USB microphone systems. It happens when USB digital signals cross over into the USB battery wires and the microphone doesn't have enough battery filtering to deal with it — typical of lower end audio and particularly microphone equipment.

It's darn fun to get rid of because the sound is exactly where your ear likes to work. The baby screaming on a jet sound.

There is no easy solution. Some combinations of computer, cable and microphone do it much worse than others and there is no magic bullet. If you talk to three different people who got relief, they all got it a different way.

Koz

[Trebor]

What would you recommend for removing this pesky ringing sound ...

Notch filters at 1KHz intervals : paste Steve's code below into a thing called Nyquist prompt , then apply it like an effect to remove the whine noise ...

Code: Select all

(let* ((q 50)                 ; set the base Q for the filter
       (freq 1000)            ; set base frequency
       ;set the number of iterations
       (iter (truncate (/ *sound-srate* (* 2 freq)))))
  (dotimes (i iter s)         ; start the DO loop
  (setf s (notch2 s (* freq (1+ i)) (* q (1+ i))))))

winkler-noise2 , before-after code shown.flac

(409.31 KiB) Downloaded 72 times

[featheredtar]

What would you recommend for removing this pesky ringing sound ...

Notch filters at 1KHz intervals : paste Steve's code below into a thing called Nyquist prompt , then apply it like an effect to remove the whine noise ...

Code: Select all

(let* ((q 50)                 ; set the base Q for the filter
       (freq 1000)            ; set base frequency
       ;set the number of iterations
       (iter (truncate (/ *sound-srate* (* 2 freq)))))
  (dotimes (i iter s)         ; start the DO loop
  (setf s (notch2 s (* freq (1+ i)) (* q (1+ i))))))

winkler-noise2 , before-after code shown.flac

Thanks, that works perfectly!

[featheredtar]

Apply:
High Pass Filter, Rolloff 24 dB, Cutoff Frequency 10 Hz (could probably use 20 Hz or even 40 Hz - higher values will remove more low-frequency noise)
Notch Filter, Frequency 60, Q 10
Notch Filter, Frequency 120, Q 10

Now apply Noise Removal:
Select a noise-only portion, then Effect > Noise Removal > Get Noise Profile
Select the entire track, then Effect > Noise Removal
- set Noise Reduction to 12 (a higher value will remove more noise at the expense of "tinkly bells" artifacts), then OK to remove the noise.

How I determined this:
I selected all of the noise-only track and did Analyze > Plot Spectrum. That showed the low-frequency noise, and the two spikes at 60 and 120 Hz. It also showed the "ringing" spikes, but there were too many to use the Notch Filter to remove them all. Getting the low frequency noise and the two low frequency spikes out of the way made it easier for Noise Removal to deal with the ringing.

-- Bill

Thanks Bill. I ended up using the Nyquist prompt code below.

[Trebor]

Apply: ... Notch Filter, Frequency 120, Q 10

Thanks Bill. I ended up using the Nyquist prompt code below.

That Nyquist code doesn't reduce the mains hum : applying a notch-filter at 120Hz (mains-hum 1st-harmonic) in this particular case does make an audible difference after the Nyquist code has been applied.