Doctor's device is cure for tone deaf

Auto-Tune can improve a singer’s pitch but it can’t take a bad singer and make them great, writes HARRY O'RAHILLY

Auto-Tune can improve a singer's pitch but it can't take a bad singer and make them great, writes HARRY O'RAHILLY

THE LUNCH guests looked down at their plates and fell silent. Even with all their technical expertise and the optimism of a good appetite, they knew they couldn’t write software to make her sing in tune.

Among the lunch-goers was Dr Andy Hildebrand, a successful engineer who at age 15 had entered the Jacksonville Symphony as a flautist and had put himself through college with music scholarships, finally earning a PhD in electrical engineering from University of Illinois.

While in college in the 70s, he began playing around with computers and synthesisers. At the same time, the music industry was getting interested in electronics and Hildebrand found that he could make money in his spare time by improving how music data was processed.

READ MORE

Hired by oil company Exxon for their research department, Hildebrand continued working in sounds and signals. Just as sound travels at different speeds in air, water or metal, an oil deposit has its own acoustic signature. By setting off dynamite underground, geologists can gather a large amount of sound data from what lies beneath the surface of the earth, thereby helping to identify suitable fields for exploration. During Hildebrand’s time at Exxon, the task of compiling, cleaning and making sense of this data was highly specialised and labour intensive.

Hildebrand felt there must be a better way so, leaving Exxon, he founded Landmark Graphics in 1982. The company’s business plan was to take the raw acoustic readings and process them into a colour coded map describing an area’s geological composition. As this automated laborious analysis, it delivered huge productivity gains to exploration companies. So valuable was the system that Landmark was acquired 13 years later for about half a billion dollars.

And then Hildebrand sat down for lunch and was asked whether he could make someone sing in tune. Despite his expertise and success with acoustics, the task seemed impossible and it was a year before he began in earnest on Auto-Tune. Hildebrand explains: “There’s not a whole lot of difference between the signal processing of vocals or seismic data . . . the problem was doing it in real time.”

The voice produces vast amounts of data per second and processing it with the chips then available was prohibitive. So how did he get Auto-Tune to work? “First of all you down-sample the data to a much lower sample rate, because that does not disturb pitch. That reduces the amount of computing. Then it’s a matter of being intelligent about what you compute and don’t compute. At that stage, you go back to the original data and apply the changes to make the voice sound like it’s being sung naturally on the right note.”

By 1997, he had a working pitch correcting device, but what would the music industry think? Hildebrand needn’t have worried. “They ripped it out of my hands.”

The studios now had a tool that could cut studio time needed to record vocals from one day, to three takes; the economies were obvious.

There were other consequences from the adoption of Auto-Tune. Hildebrand tells the story of a music producer who jokes that once he used to “look for good singers, now . . . it’s pretty faces”.

Unsurprisingly, the use of the software has drawn a lot of flak, with bands such as Death Cab for Cutie launching an Anti-Auto-Tune campaign and accusations abounding over the authenticity of shows like X-Factor. Does he feel the same way as Feynman or Einstein did after helping to develop the atomic bomb?

With a laugh he answers: “I just build the car, I don’t drive it on the wrong side of the road.

“It’s not just used for what people hear as the ‘Auto-Tune effect’ on songs; it’s used on almost all songs to fix the pitch just a little bit here and there, and it’s used in a way you can’t tell.”

In this respect it sits alongside other computerised tools used by sound engineers to amplify, cut and mix audio.

With a symphony orchestra background, it’s not surprising that he listens to classical music, but he also has more popular tastes – “ . . . there’s lots of rock and roll I’m interested in, but I think rock and roll changed about 20 years ago and I’m not interested in the recent stuff.”

With all his connections and expertise, has Hildebrand any intention of releasing something himself? “I had a dream about that the other night,” he chuckles. “I was gonna become a rap artist and speak my mind about the 99 Percenters [the campaigners occupying Wall Street]; I’m much more conservative than they are, they should get a bath and get a job, and use Auto-Tune to set this [idea] to rap music!”

While still keeping his eye on developments in seismology, he has no plans to go back. “I had some ideas for software, but when I had a look at Landmark they had already got there.”

Instead, his focus is on Solid-Tune, a newly developed gadget for guitars that provides different tunings, pitch shifts and many other kinds of effects.

While the balance struck between image and talent may have shifted, Hildebrand is clear: “Auto-Tune can make a bad singer a little better, take an average singer and make them sound good, but it can’t take a bad singer and make them sound great.”