3 replies to this topic

[skiesbleed]

Hello everyone,

I've been thinking about the topic of trying to take two tracks and superimpose them to purposely have certain frequencies cancel out. For example, a stereo song that sounds different in mono, because of calculated cancellations between the left and right tracks. Now, I know that you could completely cancel out the sounds if for every single frequency you shifted the phase by half of its period, but this poses two problems. First of all, creating band passes to split all of the frequencies is tedious, and would likely need to be done using an algorithm. Second, this only allows for the waves to be completely cancelled, and not for different sounds to emerge, unless of course those sounds were already present and the frequencies at which they exist were not phase shifted between the two audio tracks. What do people here think? Can anyone think of a means of achieving this kind of effect, besides designing a program to process the audio, or how to only mask certain parts when the tracks are combined? I'm interested to hear if anyone has thought about / worked towards something like this.

[Jester_Fu]

If you're looking for an evolving effect, you need to work out superposition using nodes. Mathematically, you don't need an algorithm to do it - just google for 'wave superposition' and you should get some pointers to more key terms on frequencies that create beating. You don't need to have standing waves (i.e. perfectly out of phase) to create nodes and antinodes between frequencies. It's all about ratio. You'll find, with a bit of reading, that there's quite a few common frequencies that will interact at certain points to increase and decrease volume of the overall sources.

Now, i can do it from first principles using maths...  but i generally do it by 'feel' in the studio hen setting up synths and i want that effect... so i can't give you a real tip as such on which frequencies to start with other than to say "seek and ye shall finde". You don't even need singular frequencies - mixed sources work well when you band pass them a little. In fact, the combination of the band pass opening/closing coupled with the beating you get from the waves interact creates some very nice effects

[rhythmboy]

I think I get what you're saying skiesbleed, but in any case you've inspired me to share a simple but effective technique that may help you. It's simply uses EQ, pan and short delay.

The underlying principle is that as you touched on, perfect cancellation occurs if the inverse waves are identical in frequency, amplitude, time position, and location in the stereo field. You get absolute cancellation if the inverse wave is a mirror-image of the original.

However, if we change any aspect of a given frequency's amplitude, time position or location in the inverse wave, even by a small amount, we no longer cancel out that frequency and it will become audible again.

Try copying a complex wave like a full song, invert the copy and run it along the original. You have perfect null/cancellation. Then on the inverse wave, insert a basic EQ that uses minimal processing (i.e. induces zero latency or sample delay) and choose a steep notch EQ with narrow bandwidth and the lowest cut level you can do.

Now sweep the EQ frequency - you will hear the precise frequency you are cutting. It's a total reverse way of thinking.

But imagine a 7 or 15 band EQ with narrow steep notches, each automated to sweep across the spectrum. Each band will reveal different elements of the song at different times. Automating the cut (-dB) amount also affects how loudly the freq's come through. It's fun 

Also try simply sweeping the pan pots around and you'll hear the mix come back in spots.

Lastly, try adding a very short delay to the inverse wave and automate the delay time to shift from zero to about 10msec and see what happens

Then you mix all the above techniques together, give it to a mastering engineer and watch him crap himself when he sees on the 'scope for the first time  ;D

[skiesbleed]

Thanks for the responses!

Thanks for the pointers, Jester_Fu. I have a bit of a physics background so I know basic wave theory, and I know about creating beats and such, but I suppose I'm more so looking for tools to do this with audio. Thank you for sharing your knowledge, though.

Rhythmboy, that is what I'm talking about, so thank you for sharing! I'll definitely need to play with the techniques you've described. I actually completely forgot that my DAW had the ability to invert waveforms, otherwise that would have answered much of my question, haha. I'm glad I asked, though; there's definitely a wealth of information and techniques here that I might not have stumbled upon for a long while yet!

Thanks again, guys.