Sound from a Virtual Planet

There’s something innately appealing to us about human sounds, the sound of a voice, hands clapping, whistling, vocal beat-boxing. Sound designers have been acutely aware of this, and have used the voice to create innumerable sound effects –the mouth is their secret weapon. One of the first classic sounds from the early days of sampling was the sound a human blowing air across the top of a glass bottle. More recently, Bjork explored the possibilities of human sounds in her haunting work Medulla, where just about all the sounds on the CD are derived from the voice or body. While it’s easy to record the sound of humans, making expressive sampled instruments from those recordings can be a daunting challenge.

This spring, SONiVOX released an interesting sample library aptly named Anatomy. Anatomy is a massive collection of 856 sampled instruments formatted for Native Instrument’s Kontakt 2 or 3. All the sounds in the collection are derived from some sound produced by the human body, and are divided into two categories: man and machine. Sounds that are distinctly human like singing and screaming are in the Man category. Those listed as Machine are sounds that have been mangled beyond recognition to create all sorts of instrument sounds, from drums and basses to pads and leads. While this is not the first time human sounds have been made available in a sample library, the sheer size of this collection is unprecedented.

While some of the instruments included here, like the vocal beat-boxing kits, are what you would expect to find, the real strength of this library is in the range of unexpected permutations that bear no resemblance to the original recorded source. Even sounds that are bizarre or aggressive have a certain appealing quality to them. SONiVOX provides some demos on their website, and here are a couple done by former students of mine from Berklee:

Anatomy Demo 0 - composed by Ben Cantil - 3,228k

Anatomy Demo 2 - composed by Stephanie Olmanni - 2,037k

To further entice you, SONiVOX offers ten of the Kontakt instruments as free downloads that will allow you to load and experiment with these on your own.

SONiVOX Anatomy Free Samples

If there is any critique of the library, it’s that you need to own Kontakt to use it. For the last few years, manufacturers have been relying on various sample players, like Kontakt Player used in the Garritan Personal Orchestra, to deliver programmed content. When you buy a library, the player is included. This has the distinct advantage of insuring compatibility, and eliminates the need for manufacturers to develop the same library for multiple sampler formats. However, for anyone interested in developing their sound design chops, there’s a big advantage to delivering the library as Kontakt instruments. You can open any of the instruments in Kontakt and see exactly how it was programmed; educational opportunities abound with examples of both standard and more exotic programming techniques.

For some time we’ve taken pitch transposition and time-stretching of audio for granted, with one caveat; only an entire audio signal gets processed. We’ve come to expect programs like Ableton Live and now Pro Tools to treat audio like butter, speeding up and slowing down loops, time correcting sloppy drumming and pitch correcting the wanna-be diva. The Waterloo of this technology has traditionally been isolating events within an audio file. If a singer hits the wrong note, it’s usually not a problem to correct, since it’s a single monophonic event that can be easily isolated and processed. What about when the piano player hits the wrong note in a chord? Well, that’s another take…. We’ve always operated under this assumption, and it was a great reason to record performances as MIDI data, since we had the freedom to freely manipulate individual notes a chord.

All this changed last week at Musikmesse, Germany’s massive musical instrument trade show, when Celemony debuted their Direct Note Access (DNA) technology. DNA can analyze an audio event, isolate individual pitched elements, and freely manipulate them in pitch and time. While I’ve said that changes in music technology products are often evolutionary, not revolutionary, this is a really big one.

Celemony started in 2001 with the initial release of Melodyne, the brainchild of German programming whiz Peter Neubäcker. The whole idea behind the technology was to allow users to edit the pitch and timing of a note graphically. Melodyne does this by analyzing the source and displaying the result as graphic data on a pitch and time grid. From here, the audio properties can be manipulated much like note and controller data in a MIDI sequencer’s graphic editing window. Opcode first introduced this concept in the late 1990’s with their StudioVision sequencer. Here, monophonic audio performances were analyzed and represented as MIDI note and pitchbend data. You simply edited the MIDI data, and rendered the result back to an audio file. Melodyne expands on this, bypassing MIDI altogether and greatly enhancing the resulting sound quality.

Melodyne is available in a variety of products from the entry level Melodyne Uno to the flagship Melodyne Studio. Starting in Fall 2008 with version 2 of the Melodyne plug-in, Direct Note Access will be incorporated into their full line and perhaps inspire some interesting new products.

Peter Neubäcker freely admits that he assumed extending the Melodyne model to individual note events in a chord was not possible. Only after challenging this basic assumption did the algorithms behind this begin to take shape. The DNA acronym works, since Direct Note Access is really about exploring the genome of the harmonic life of music. While the demo video is truly amazing, keep in mind that the types of performances here illustrate what may be the best case scenario for effectively using this technology. There are limits, and one wonders if DNA could root out a wrong note in a dense, orchestral recording. Still, what Celemony has come up with is nothing short of remarkable.