Craig Anderton’s Open Channel: Strip-Mining the Emotion From Music

There’s an uneasy feeling that something is not quite right with today’s music—and it goes beyond the boilerplate “kids, get off my lawn” you expect from older generations. Why does a younger demographic believe that vinyl sounds better? Why did the 23-year-old server at a local restaurant tell me she preferred 1960s and ’70s music? Why do kids barely into their teens wear AC/DC and Pink Floyd t-shirts?

Today’s listeners have access to most of the music catalog at their fingertips and they stream music nonstop. But often, it’s background music. Ask anyone about the last time they actually just listened to music, at the exclusion of other activities. Expect a long pause as they try to remember. Something’s wrong. And I think I know what part of it is.

With the advent of digital audio, sound quality started to be confused with musical quality. We can make recordings that sound absolutely stunning, but are musically bland. The music may not have started out that way, but in the wrong hands, technology intended to improve sound quality can strip the emotional impact inherent in music’s crucial expressive elements: timing, dynamics, pitch and tension/release.

Consider pitch correction. Most people who don’t like it cite the unnatural sound, or feel it’s “cheating.” But I think a bigger problem might be that the human voice is not a fretted or keyboard instrument.

A great vocalist will slide around notes and play with pitch. I recently analyzed a vocal line that I thought sounded excellent. It started with a series of five notes, going from a low note based on the tonic, to a high note. The low and high notes were right on pitch. But the three in-between notes were not only “incorrectly” tuned, they were off-pitch in what seemed like a deliberate (albeit probably subconscious) way.

The first of the three notes was considerably flat. The second of the three was a little less flat, and the third note was very slightly flat. This seemed like a classic example of building tension for tension-and-release. The flat notes caused tension, and release happened after coming ever-closer to, then ultimately hitting, the perfect high note. Processing the vocalist’s pitch would have corrected the tuning, but strip-mined the emotion behind the pitch.

Our ears pay attention to pitch. When you’re not sure about something and say, “Uh-oh,” your voice drops in pitch compared to your normal speaking voice. But when you say, “Look out,” the pitch rises. It also rises when you ask a question, but falls to indicate the end of a thought. In tonal languages, like Mandarin, Vietnamese, Cherokee, Zulu and others, the same word can have multiple meanings depending on its pitch variations.

Craig Anderton’s Open Channel: The AI Cesspool Has Arrived

The fundamental flaw in pitch correction is relating pitch solely to a technical model—the irrational number-based intervals of an even-tempered scale—rather than a musical model. Singers often glide to notes based on just intonation, even with a band playing even-tempered instruments. Yet it doesn’t sound out of tune. Instead, the subtle pitch differences enrich the sound—like what happens when instruments have doubled strings, or use chorusing effects. String ensembles and barbershop quartets instinctively see pitch not as a technical issue to be conquered, but an emotional concept to be understood.

Bob Dylan emphasized notes by singing sharp. Madonna left in a clearly out-of-tune slide at the end of “Ray of Light,” which made the song just that much more compelling. When Judy Garland sang “somewhere over the rainbow,” the pitch of “somewhere” followed the upward visual arc of a rainbow, while “over” slid down the other side of the rainbow’s arc. When Robert Plant sang “way down inside” on “Whole Lotta Love,” he used pitch to bring “way” up for emphasis, only to slide down with “inside.” His use of pitch was mathematically incorrect—but, musically speaking, it was perfect pitch.

We’re at the column plot point where it’s time to offer some proposed solutions, so here we go:

• Correct only notes that sound wrong. It doesn’t matter whether they look right or wrong.
• Don’t let engineers decide when to use pitch correction (no offense, y’all).
• Have the singer listen to the track. If the singer hears something wrong, then the engineer fixes it.
• Alter pitch in context with the rest of the music. Variable-level dim solo, not standard solo, is your friend.
• Always turn off snap/quantize-to-scale with any pitch correction software. Instead, if a note sounds “wrong,” grab the note while monitoring its pitch, and move it until it sounds right. Don’t be distracted by any tuning calibrations. They’re not listening to the music.
• Use pitch un-correction. As with the example of the three notes that ended on a note with perfect pitch, you can introduce tension-and-release elements by altering pitch strategically to make emotional statements. Go a little sharp on the last held note before the big chorus to add friction. Go slightly flat with a note that’s just before the end of a phrase, and end on a perfectly tuned note. Also, harmonies complement a lead vocal well with judicious use of slightly flat or sharp notes.

Yes, it’s fashionable to complain about pitch correction. And, granted, I’m complaining about correction of pitch. But I’ll never complain about freedom of pitch. To me, that’s where pitch-altering tools excel—they give the opportunity to enhance a song’s emotional impact, not strip-mine it.