In the Mac OS X world, things have been busy at MyDreamApp, the brainchild of a few elite developers. In American Idol fashion, users were encouraged to submit their ideas for their favorite application ideas. 24 finalists were selected, with progressing rounds of elimination based on open voting. The top three apps that win will actually be developed full-scale and sold for market consumption.

Of the entries, the most intriguing (and the only music-related) concept is Whistler, the brainchild of live music photographer Richard Whitelock. From his submission description:

Ever had the urge to create a song until you realized it was harder than it was worth? With Whistler, just whistle, hum or tap out your creation into music app importable form.

The idea has merit, many electronic artists in particular employ a varying number of digital sketchpads to quickly jot down ideas and Whistler would fit neatly into that toolbox. But the nature of what the program intends to do is far more ambitious at the core than imagined by Richard or the developers who selected it as a finalist, not quite realizing the potential for a full-blown white elephant.

Next up, Richard’s latest mock-up visualization of the program’s GUI and functions:

It’s hot, no doubt. To be able to tap your fingers on the desk or notebook trackpad, to whistle or hum a tune all embodied in a lightweight, easy-to-use Cocoa interface would be some of the slickest execution in modern application development. Certainly it would attract legions of new composers to the art of digital production, a veritable first-step into the black hole that swallows more free time from DAW-based musicians than any of them likely care to admit.

But therein lies the rub. Transforming raw audio in to sheet music is far easier said than done. Logic Pro offers both pitch detection and groove templates to sort out the fundamental concepts behind Whistler, but what it seeks to accomplish is leaps and bounds beyond the technical feasibility displayed in most commonly used music applications. Danny Patterson (A resident synthesis major at Berklee School of Music) correctly pointed this out.

You can do that with Max/MSP or something. There’s no way to make it a cocoa app easily. Auto chords don’t make sense. You’re in a key, and each note has a different chord associated with it (or number of chords. But even if every note has the same chord it’s not musically correct because chord tones of one chord are not chord tones of another note in that scale

Richard maintains his focus has always been on converting variable-range input into monophonic MIDI hits for sequencing, but that idea simply serves no additional purpose besides getting a glorified paint-by-numbers monophonic MIDI sequencer that you’d map to a virtual drum kit in GarageBand. Which could be fun, but not all that flexible in the grand scheme of things. However, by making a leap into polyphonic recording, Whistler would no longer be a curious surface-scratcher for beginners, but a legendary contribution to digital composition. Given Richard’s vision thus far, this would also introduce the most simplistic, easy-to-use interface of any audio application on the market.

It is seriously worth exploring the technical feasibility of converting polyphonic MIDI signals on the fly. You suddenly have a full-blown digital tuner in your hands, capable of tuning strings on any instrument from guitars to cellos, training high school trumpet players, keeping practicing singers on their notes or detecting the pitch and key of any instrument available. Such a capability masked under an intuitive and simplistic interface would be a revolutionary breakthrough.

There are a handful of programs that already do this, but they’re complicated and priced for a niche market. Max/MSP and Melodyne offer variable approaches to pitch detection in addition to built-in functions in Logic Pro and MOTU’s Digital Performer. Berklee professor/performer Neil Leonard explained how he tunes his bass clarinet with instant, real-time processing:

I use Max/MSP with Fiddle [a Max/MSP External] for pitch following. Before that, I used the IVL pitchrider and the Roland CP-40.

The technology is all here, it’s been done before. The trick is implementing it in a way that allows beginners and experts alike tap the potential offered. Additionally, you could set a key as C major for example, and round the inputted notes according to the selected scale. It’s an adventurous concept, but worth implementing as it would actually train the user to stay within a key instead of randomly bouncing around like an undisciplined, homebrew Aphex Twin.

The bottom line is, this is very doable and offers a hugely viable concept commercially if successful. And even if this application fails the contest, it’s a worthy project for an Objective C programmer to pick up and put together. The potential for utility and entertainment both are immense, and if executed properly would be established as one of the legendary few applications to revolutionize the user experience on Mac OS X.