August 3, 2011 at 1:52 pm

Spectral Layers, No Longer Vaporware, Edits Music ‘Like Photoshop’ (Updated)

It’s still in an unreleased “vaporware” stage (update: not anymore), but nonetheless we’re intrigued by Spectral Layers, described as “a kind of Photoshop for sound” by its developer, Divide Frame. In more scientific terms, what it does could be described as spectral analysis and re-synthesis.

This Mac/Windows desktop software is a little outside the wheelhouse, but it’s just so pretty, we couldn’t resist posting the above video. Ostensibly, this technology should eventually allow us to fingerpaint with sound on our tablets in a whole new way.

divide frame spectral layers edit audio like photoshopIn a series of tutorial videos by the developer, Spectral Layers performs a host of neat tricks on a variety of sound files. It separates speech from background music; unwanted noise from a poor-quality recording; individual instruments from a duet. Impressive.

The user appears to perform these tasks quickly and easily, by editing a spectrogram image — essentially a visual breakdown of a sound into its constituent frequencies (we saw a similar approach from Celemony Melodyne). A squiggle near the bottom of a spectrogram might represent the bass line of a rock song, while some at the top could be the crash of the drummer’s cymbals. With Spectral Layers, the user can select these elements individually and manipulate them in any number of ways.

Do you like Led Zeppelin but hate the otherworldly screech of Robert Plant? No problem — just find the frequencies that make up his voice and transpose them down into a more palatable range, leaving the rest of the instruments intact. Or if that’s not enough, remove his voice altogether to create a karaoke backing track. Feeling a little more avant-garde? Stretch, transpose, and generally mess around with all the frequencies of a sound until “The Song Remains the Same” sounds more like “Gesang der Junglinge.”

These are the sorts of things made possible by spectral analysis and re-synthesis — the process that is ostensibly at the core of what Spectral Layers does. It works by analyzing the frequency makeup of a recording in order to rebuild it from scratch out of lots and lots of sine waves (the basic building blocks of sound). This reverse-engineered version of the sound, generated by the program, gives the user much greater control over individual frequencies than he/she would have trying to manipulate the original recording.

While this all may sound a little familiar to users of Michael Klingbeil’s SPEAR – a freeware spectral analysis/resynthesis program that’s been around for a while — there are a few features that might give Spectral Layers a leg up, assuming Divide Frame ever releases it to the public (update: this software is now available; it’s expensive, but there’s a free trial).

These similarities include some novel ways of dealing with stereo files, (something SPEAR can’t do) a better-looking interface, and automatic detection of harmonic spectra, which should simplify the isolation of individual instruments within a mix.

There’s no word yet on a release date for Spectral Layers. But if the average user can get anything like the kinds of results shown in the videos — especially through a more beginner-friendly tablet or smartphone interface — it will almost certainly be worth the wait.