Adding a kazoo effect

Audacity 2.0.2, Windows 7x64

This might seem like a strange request, but I teach English as a second language and one of the best methods to teach rhythm and intonation is to hum the words through a kazoo. This allows the student to hear the pattern without being distracted by the words. I was thinking that if I could distort a speech recording in a similar pattern, it would be a great help to the students as they could take the recording home and practice.

Since I am a complete beginner with Audacity, I would greatly appreciate any help I could get in how to set an effect like this.


I can’t think of a way to do that. Anything that will eliminate the pronunciation of the words enough to obliterate them will also take out the tonal variations you want to preserve. You can trash human speech remarkably and still be able to understand it – see cellphone calls.

I’m not completely clear on who is going to do the kazooing you or the students. Or both.


You did give me a terrific idea. People are always posting about applying the “announcer filter” and make them into a famous announcer/presenter. Of course, there is no such thing, but we always get stuck on why. How do you explain vocal acting, interpretation, and expression.

With a kazoo.

Let me work on this.


As a side note, did you see the movie “Despicable me?” The little yellow minion characters are famous for not using any known language, certainly not English, but they have zero trouble telegraphing their meaning with whatever sounds they make. That’s a cousin to what you’re doing, right? Koz

Would that work?

That’s an existing voice I had for another project.

Effect > Low Pass Filter, 36dB, 0.7071, 300Hz
Effect > Amplify > OK
Effect > Leveler, Heaviest, -70dB > OK
Effect > Leveler, Heaviest, -70dB > OK

If you mess with the Low Pass Filter “300” number, you can get it a little more muffled. Try 250 or lower. You can experiment and see what happens to your voice.


This sounded like fun, so I’ve written an effect plug-in.

I would say a sentence into the kazoo and the students would pick up on the the stress, intonation and rhythm. Contrary to what most people think, good pronunciation is not mostly about saying “th” or “r” or “l” properly.

I once asked a Chinese student what sports he likes. He replied," bad-MIN-ton" (with huge stress on the second syllable-try it yourself). For the life of me, I could not figure out out what he meant. After he repeated it about 10 times, I finally figured it out.

Thanks, that’s exactly what I meant! I’ll try it out plus the plug-in from Steve.

I’ve just updated the “kazoo” plug-in so that it works with stereo tracks. (the first version was for mono only).

I think I make a much better mess than kazoo default does.

I posted the original “clean” voice track.

And the default kazoo Nyquist.

Like I said, it’s harder to mess up a human voice than you would think. I’ll try higher filter values.


No. The packaged filter doesn’t do near enough damage. Koz

You can do a bit more damage with the new version: Kazoo effect - #6 by steve

similar question … Removing words from speech, and maintaining the tone

envelope follower” is another possibility, but the result would be monotone, (whereas kazoo can vary in pitch)

Yes, we shouldn’t wander too far away from the goal: an intelligibility notch filter. Rhythm, overall pitch, interpretation, swing, interstitial and emotion and a complete inability to tell what I said.

Also see: The grownups talking in any Charlie Brown animation, carried, unless I miss my guess, by a trombone performance.


Forgive my ignorance as I’ve just started using Auadacity, but what’s the 0.7071 in the Low Pass Filter? I just get 2 fields–rolloff and cutoff frequency? And I’m assuming that you apply the Leveler twice?

That’s pretty close, but I can still make out the words which wouldn’t be ideal for pronunciation training.

Older versions of Audacity had a third control for setting “q” (the filter quality). That control has now been removed and the effect automatically sets the optimum “q” value.

So what exactly do you want to be able to hear? Just the stress and rhythm of the words but not the words themselves?
If so, try entering this code in the “Nyquist Prompt” effect. (If this does what you want then I can turn it into a plug-in to make it easier for your students to use)

(defun env (sig)
  (let ((step (truncate (+ 0.5 (/ (snd-srate sig) 200)))))
    (mult (osc (hz-to-step 200) 1 *tri-table*)
      (snd-avg (snd-abs sig) step step op-peak))))

(multichan-expand #' env s)

Exactly. I don’t want them to hear the words because then they focus on the meaning and whether they are pronouncing the individual letters correctly instead of focusing on the stress, intonation and rhythm.

As I mentioned previously, I’m a complete beginner with Audacity, so I really don’t know what to do with the code you provided.

  1. Select the audio track (as you would for applying any other effect.
  2. Effect menu > Nyquist Prompt
  3. Copy and paste the code into the text box.
  4. Click the “OK” button.

Yes, I applied the filter twice. Once wasn’t enough damage. Koz