How to convert a voice to just a tone to reveal the pitch?

I need to help some students to improve their awareness of the pitch changes in intonation. I’ve seen some teachers use a slide whistle to show this.

I have some audio samples of speakers reaching some sentences naturally recorded professionally, so there is no background noise. Is there any way to convert the audio to something that sounds like a slide whistle or a constant beep like one hears in a hearing test? I want the students to not hear the words, but to just hear the changing pitches.

It doesn’t seem like it, but this is dangerously close to the frequent request: “How do I make my voice sound like something or somebody else?” This is usually in the context of “make me sound like a professional announcer,” but it can also be a way to simulate a game effect or sound. “Make me sound like a Mega Zorroid Master in my favorite game.”

Oddly, this isn’t all that hard to do backwards. Effect > Vocoder can give you the Talking Guitar effect, where the pitch comes from the guitar but the intelligence comes from the actor.

You may get close by choosing good values of Effect > Low Pass Filter. Most intelligence happens in the 1000 to 3000 frequency range. This is the “telephone” effect and that’s why those frequencies were picked. They 're the minimum needed to make out what the person at the other end is saying.

So try values of 1000, 2000 and 3000 in Low Pass Filter and see if any of those results will work for you.


Okay, I’ve tried several attempts with the vocoder filter and low pass filter. I set the low pass filter to 3000, 2000, 1000, 500, 20. At lower numbers, the sound is quieter, but the words are still clearly recognizable.

All you need is the low pass filter. Forget the vocoder.

As you pass 3000 on the way down in experimental steps, the words should be harder and harder to understand. Pick a higher rolloff than the default 6. Go for 18.

15000 is sparkly, clear FM radio, 5000 is AM radio, 3000 is telephone quality. Voices become more and more muffled and harder to understand as you go down.

The problem with this technique is eventually, you’ll filter out the exact tones you’re trying to preserve.

You may not get as close as you need.


A pitch tracking VST, e.g. Kazoo* , or Alien-solo* … Charlie Brown (Peanuts) Teacher Voice - #6 by tedbear [ * those are Windows only ].

“sine speech” fits that description …

A human voice is comprised of a broad range of complex frequency patterns. I think what you are wanting to do is to try and isolate the dominant frequency bands, which means silencing all of the “lesser” frequencies. To do this completely is difficult, but you can sometimes get close to the effect by abusing the “Noise Reduction” effect.
Here is an example:

This was created by using the entire voice recording as the “noise sample”, and then applying the Noise Reduction effect with settings of 48, 10, 0.

I forgot about the amazing DTBLKFX plugin which can mangle sound like nothing else …

squeaky-gate.xml (503 Bytes)