Hiding mono audio within stereo audio - help me figure this out

Ok, so help me figure this out. In this clip Mick Gordon explains (poorly) how he hid a voice sample within a drone piece.
 

I've googled it and found a couple of people talking about it, but this explanation is such a mess.

I've also asked ChatGPT to help me figure it out, but the answer made no sense and obviously didn't work.

In the end, all I want to do is add a little easter in something I'm working on using the same technique.

did you try it? 

basically he's saying in stereo no one will notice but if someone listens in mono they hear it.. so unless someone has a mono listening environment they won't hear it. 

do you have a way to monitor in mono? "sum to mono" button somewhere like on a monitor controller or via audio interface? 

when you take mono sound and put it on 2 tracks (dual mono) and invert (flip) the phase of one channel they will cancel out and often if you flip the phase of one side of a stereo signal you will get some cancellation of frequencies depend on how it was recorded etc... so, he's cancelling the signal using phase flipping to reveal the mono voice buried in the track.  but this only works when someone listens in mono. 

I made a track with brownish noise on one channel, using a stereo shaper to route L to L and inverted L to R on that channel only, with some clap sounds on another unaltered channel, lower in volume so you cant hear it amidst the noise (noise must contain loud enough of similar frequencies to hide it), and then rendered in stereo and mono formats:

however it does mean that that section of the track has to be effectively mono - like if you're hiding it in music it might be noticable if it switches to mono, so you might have to play around with the level of the hidden part and the mono vs stereoness of the main sound - I dunno if there's a way to do the same with a fully stereo track, I can't think of one

Dunno if this makes it more or less clear lol: 

  On 3/25/2025 at 1:16 PM, ignatius said:

did you try it? 

basically he's saying in stereo no one will notice but if someone listens in mono they hear it.. so unless someone has a mono listening environment they won't hear it. 

do you have a way to monitor in mono? "sum to mono" button somewhere like on a monitor controller or via audio interface? 

when you take mono sound and put it on 2 tracks (dual mono) and invert (flip) the phase of one channel they will cancel out and often if you flip the phase of one side of a stereo signal you will get some cancellation of frequencies depend on how it was recorded etc... so, he's cancelling the signal using phase flipping to reveal the mono voice buried in the track.  but this only works when someone listens in mono. 

 

Expand  

Of course!
And I understand it, but for some reason I haven't been able to get it to work.

  On 3/25/2025 at 1:27 PM, hoggy said:

I made a track with brownish noise on one channel, using a stereo shaper to route L to L and inverted L to R on that channel only, with some clap sounds on another unaltered channel, lower in volume so you cant hear it amidst the noise (noise must contain loud enough of similar frequencies to hide it), and then rendered in stereo and mono formats:

 

Expand  

Yup, this is exactly what I've done. I wonder if Logic is fucked in some way. I'll try Ableton or Reaper.

I don't know with Logic but with FL there are different mono render options, like L to mono, R to mono, mix to mono etc.?

Wouldn't that depend on how the mono signal is summed though? Because if you smash the two reversed phases on top of eachother it surely would cancel out the sound, no? If you only output one of the channels you'd probably hear it. Am I going mad? 

Nah, that's not it. The more I think about it the less sense it makes.

  On 3/25/2025 at 2:17 PM, Silent Member said:

Wouldn't that depend on how the mono signal is summed though? Because if you smash the two reversed phases on top of eachother it surely would cancel out the sound, no? If you only output one of the channels you'd probably hear it. Am I going mad? 

 

Nah, that's not it. The more I think about it the less sense it makes.

You're making the sound disappear which covers the hidden sound - like in those bank letters with your pin number that have a plastic window covered in random bits of numbers over the top of the real number, so you can't see the real number through the envelope, and removing the plastic window is like removing the big sound covering the hidden bit

If you have a stereo track with a mono signal panned right and the out of phase version of it panned left, listening in (summed) mono won't reveal the hidden signal. Ideally, you have to listen to either the left or the right channel to hear it.

https://en.wikipedia.org/wiki/Out_of_Phase_Stereo

This is also related to mid/side processing.

  On 3/25/2025 at 3:08 PM, psn said:

If you have a stereo track with a mono signal panned right and the out of phase version of it panned left, listening in (summed) mono won't reveal the hidden signal. Ideally, you have to listen to either the left or the right channel to hear it.

https://en.wikipedia.org/wiki/Out_of_Phase_Stereo

This is also related to mid/side processing.

by summed mono you just mean adding the two sides together? if so, if it's 180degrees out of phase, i.e. inverted, that would just make it disappear, and you want the cover to disappear so the hidden signal is revealed... I dunno what I'm missing but it all seems pretty simple to me, maybe I'm dumb and you know something I don't

This is what I put on the bit I want to disappear, so the thing lower in the mix can be heard, I'm sure there's an equivalent thing in superior DAWs (hehe):

 

Like literally, look, go into your audio accessibility settings on windows and set mono to on, play the first of the two clips I uploaded earlier, then turn it back off and play it again - I just tested it, it works

  On 3/25/2025 at 2:17 PM, Silent Member said:

Wouldn't that depend on how the mono signal is summed though? Because if you smash the two reversed phases on top of eachother it surely would cancel out the sound, no? If you only output one of the channels you'd probably hear it. Am I going mad? 

Yeah you're exactly right - though it's the music/louder thing you're hiding the easter egg with, that you have to put as duplicate mono in the left and right channels (with one of the channels phase inverted)

a) have some loud noise or music that's mono

b) put it in the left channel

c) put an inverted copy on the right channel

d) put the also mono audio you want to hide in the center, both channels, at a low level

to 'decode' just mix both channels down into mono. everything disappears completely except for the hidden audio.

we are the mono makers

It's a nice idea but I assume the hidden sound won't survive lossy compression?

  On 3/25/2025 at 5:53 PM, zkom said:

 

It's a nice idea but I assume the hidden sound won't survive lossy compression?

Probably depends on how 'low level' the sound is recorded on how much is lost during compression - it definitely would degrade it for sure, both as a result of the 'removal of less audible sounds' part of audio compression and will no doubt affect the phase information of the two channels

I tried this in bitwig but the result wasn't very compelling. Is the original file somewhere out there so it can be dissected to confirm this is whats going on? I bet it works better when it's not white noise, but some other kind of drone thats masking a similar frequency range. Actually in mick's example the drone is more tonal and the vocal is a whispery noisy vocal bit. So I inverted the exercise in a sense.

  On 3/25/2025 at 6:34 PM, Bubba69 said:

I tried this in bitwig but the result wasn't very compelling. Is the original file somewhere out there so it can be dissected to confirm this is whats going on? I bet it works better when it's not white noise, but some other kind of drone thats masking a similar frequency range. Actually in mick's example the drone is more tonal and the vocal is a whispery noisy vocal bit. So I inverted the exercise in a sense.

Expand  

Brown noise is better on average because higher frequencies sound louder at the same amplitude than low ones, but it depends what the hidden sound is

  On 3/25/2025 at 5:53 PM, zkom said:

It's a nice idea but I assume the hidden sound won't survive lossy compression?

The example I posted earlier was 192kbps MP3 and it seemed fine, have a listen to it - it wasn't a very complicated sound though

  On 3/25/2025 at 7:03 PM, hoggy said:

Brown noise is better on average because higher frequencies sound louder at the same amplitude than low ones, but it depends what the hidden sound is

The example I posted earlier was 192kbps MP3 and it seemed fine, have a listen to it - it wasn't a very complicated sound though

Oh yeah, it's surprisingly clear. Interesting because I would have assumed a lot of the phase information would have been lost or gone all wobbly.

I did another one.. it has to be kinda quiet in practice, depending on the tracks - the hidden sound was recorded on my phone so its a bit scratchy