A view of the world from my own unique perspective

In some of my previous blog posts, I mentioned that I like to experiment with music. I’ve created a series of music and drumbeat montages, and I’ve also been able to uncover hidden music and lyrics in pop songs.

I’d now like to share with you, a serendipitous discovery I made a while ago. While I was playing around with my audio editing software, I managed to uncover previously unheard vocals in some pop songs. To be fair, most songs yielded nothing of any interest, but one tune in particular was remarkable – it was almost as if I had discovered a secret “third channel” hidden in the stereo recording.

As we all know, almost all modern music is recorded in stereo. There was a time during the early 1970s when quadraphonic sound looked like it would become popular, but it never gained widespread acceptance. More recently, the SACD (Super Audio Compact Disc) and DVD-A formats promised music in 5.1-channel surround sound, but they never became popular either.

The source materials I used were neither quadraphonic nor surround. They were simply ordinary CDs, recorded in stereo. One afternoon, as I was extracting songs from CDs, and then playing around with the waveforms, I heard what sounded like brand new vocals and unexpected instrumentation on a 1977 song called Long Time, by Boston. What was going on here? Where was this new material coming from?

Here is a short excerpt from the song. This is the original, unaltered version:


This is the mysterious “third track” that I had somehow uncovered:


I’ve been listening to this song for decades, yet I’ve never heard these vocal harmonies before. I then listened to both the left and right channels of the original recording, but I still couldn’t hear these new vocal harmonies. I wasn’t sure exactly how these vocals suddenly became isolated and audible, but I knew that this was significant. I quickly wrote down everything I had done, so that I wouldn’t forget the procedure. These are the steps:

  1. Isolate one channel, and invert the waveform. In this case, I inverted the left channel. These are the before and after images of the waveform.

Original Wave Sample

Inverted Wave Sample

2. Combine the inverted left channel with the normal right channel, into a mono signal. This is what it looks like:

Inverted L + Original R

Naturally, the next thing I tried to do was identify and extract a fourth channel. Who knows how much more material was hidden in these stereo recordings, just waiting to be uncovered? I was convinced that I was on to something big – this could mean the rebirth of quadraphonic sound. However, unlike the early 1970s quadraphonic recordings, which had to be recorded in four channels, I would be able to extract channels three and four from ordinary stereo recordings. Imagine experiencing your existing music collection again, in four-channel sound! I was already imagining an audio renaissance and the re-emergence of the audiophile.

Since I inverted the left channel and then blended it with the unaltered right channel, I now tried inverting the right channel and mixing it with the original left channel. Unfortunately, the result was a little disappointing – my “fourth channel” sounded exactly the same as the third channel. I repeated the procedure on a number of different songs, but I wasn’t able to uncover any new musical or vocal material.

However, I did manage to create a decent third track, and even though the bass was a little weak (the wave inversion attenuates the sound, especially at lower frequencies), it could conceivably be used as a rear channel in a surround sound living room arrangement, like this:

Living Room Speaker Arrangement

This couldn’t have been accomplished in the 1970s, but today’s computers have more than enough processing power to encode and process music in real-time. As a song plays, a CPU can extract a third channel and play it simultaneously in the rear speaker, which makes this living room scenario viable.


What I Had Actually Discovered

It seemed unlikely that I would be the first person to discover this hidden material – surely, some audio engineer or hobbyist must have done some similar experiments years ago – so I decided to do a little online research. As it turned out, I didn’t discover anything new. I had stumbled upon an audio technique that is used by karaoke machines, known as the “vocal cut” algorithm. Since instrumental versions of pop songs are not always available, karaoke machines will perform the same waveform inversion and mono conversion that I did, which often reduces the volume of the lead vocals.

I didn’t pick up on this because my results were mixed – sometimes the vocals sounded a bit muffled, and sometimes they merely became very echo-y. However, this did explain why the vocals disappeared entirely from one section of the Boston song;

Original version, with vocals:


Processed version, with vocals missing:


Despite not discovering anything groundbreaking, I still think that that my rear channel living room arrangement idea is a good one. Have the left and right channels at the front of the room, but add a karaoke channel in the background, behind the listener. While my results so far have been hit and miss, occasionally – in the case of Boston’s Long Time – it will sound like new vocals are being sung from another area of the room, which will place the listener in the middle of a virtual sound stage. And this experience – giving the listener the feeling that s/he is sitting in the middle of the stage, surrounded by the musicians – is what stereo equipment manufacturers have been trying to recreate since the 1970s.




Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: