I can't recall a single system crash.
When you have a product that EM editors already billedas “the most powerful sound-design workstation on theplanet,” what do you do for an encore? Symbolic Sound, makerof the Kyma System, faced that challenge. With the recent releaseof Kyma 5.11, Symbolic Sound found a way to make the system morepowerful, easier to use, and less expensive. That combinationshould make Kyma appeal to a large number of people.
Among the most important additions in this release are amultitrack timeline for scheduling sounds and processes, enhancedcustomization options, and a slew of new sound modules. Kyma'sprototype algorithms are also better organized, and the new SoundBrowser provides easier access to project files. A number of newtools target dance and remix artists. All in all, there is much tolike in this significantly enhanced new version.
Because we have looked at Kyma several times (see the review inthe January 1998 issue and the 2001 Editors' Choice Awards in theJanuary 2001 issue), I'll give a short overview of the systembefore diving in to the new features. To hear some examples of whatKyma can do, check out www.symbolicsound.com/hearkyma.html.
WHO, WHAT, WHERE
The Kyma System combines a DSP-based “mainframe”called the Capybara Sound Computation Engine (see Fig. 1),with Kyma, an elegant graphical sound-design language. The Capybarais a 3U rack-mount unit that contains four Motorola DSP 56309s and96 MB RAM and ships with four 24-bit, 100 kHz A/D/A converters(expandable to eight). It supports an additional 2 to 12 expansioncards ($595 per card), each with two more Motorola DSP chips and 48MB RAM. A PCI card connects the Capybara to your computer. (A PCCard interface is available for connecting the unit to alaptop.)
The Capybara provides the raw horsepower needed to generate thesounds and processes you design on your Mac or PC while runningKyma. In effect, the system is a hardware-accelerated virtualsynthesizer that employs dozens of sound-synthesis methods, effectsprocesses, and a lot more.
As with a soft synthesizer, you design sounds of any complexityusing the hundreds of generating or processing algorithms Kymaprovides, then download them to the Capybara for instantaneousplayback. What moves Kyma well beyond the modular-synthesizerworld, however, are the enormous range of functions the systemoffers, the total flexibility with which you can employ them, andthe tremendous processing power at your disposal. More than 300customizable algorithms, called Prototypes, can be combined innearly any way, and there are no limits on the number of algorithmsor the manner in which they can be arranged into a finished design.You can assign Program Changes to switch among sounds, use nearlyany type of information to control a sound's parameters, and, ofcourse, build vast libraries of your custom patches.
But you don't need to build your own sounds — more than1,000 factory examples in a wide range of categories ship with thesystem and will keep you busy for hours. The examples have settingsthat make them useful right out of the box, and in short orderyou'll have your own collection of favorites.
WHAT'S IN A NAME
Kyma's naming conventions can cause some confusion. For example,all synthesis and processing algorithms are called Sounds. ThoseSounds can be anything from a process to read a file from disk (theDiskPlayer), a filter (the 2-FormantElement), one of many synthesismethods (including FM, AM, and subtractive), or even a livemultitrack looper (the Four-track looper). Individual Sounds can becombined into larger structures, also called Sounds. To hear aSound, compile and load it into the Capybara, then send it anyneeded real-time control information, such as MIDI messages or liveaudio input. Most Sounds can be tweaked so that they play back withno user input — in most cases, you can tell the Sound to readdata from a preexisting MIDI file on your hard drive or use ascript to generate the control data it needs.
To build new Sounds, open a Sound file and drag Prototypes fromthe Prototype strip into it (see Fig. 2). Prototypes arethe basic building blocks in Kyma and are not editable; when youdrag a Prototype to a Sound file, you always work on a copy of theoriginal. There are numerous shortcuts for creating Sounds, such asMIDI Learn, in which moving a physical controller attached to Kymaassigns that controller's number as the value to control a selectedparameter. To use one Prototype (such as an oscillator) to controlanother's parameter, just copy the first Prototype and paste itinto the desired parameter field of the second Prototype, andyou're set.
Kyma uses the term Hot parameters to describe thoseparameters of a Sound that can be updated in real time. Nearlyevery setting of a Sound can be a Hot parameter, and when you firstcompile and load a Sound, a Virtual Control Surface (VCS)containing sliders and knobs for all the Hot parameters anywhere inyour design appears on the screen (see Fig. 3).
You can save VCS presets, which are snapshots of parametersettings; associate a group of settings as the starting values whenthe Sound is run the next time; and even use a random-settingsgenerator (called “rolling the dice”) to experimentwith new settings. Any parameter can be locked out of therandomization process; that option is very useful for avoidingamplitude levels that pulverize your speakers. A robust VCS editorlets you change the type of interface elements that controlparameters or rearrange their layout on the screen.
One major new feature in Kyma 5.11 is the Sound Browser. Likethe older File Organizer, the Sound Browser provides analphabetically sorted view of the files on your hard drive thatKyma can use (see Fig. 4). But now you can auditionanything from raw samples to complex, multilayered effects directlyfrom within the Browser. You can even have separate Browsers fordifferent projects or limit the Browser display to certain filetypes, drives, or directories.
The Sound Browser also gives you access to the new ReplaceableInput (RI) feature. The RI feature allows you to specify a sampleor audio file that will be used by default as the basis for anySounds that use preexisting audio.
For example, if you do sound design for a game and want to tryout numerous effects on a vocal sample, just assign that sample asthe RI, scroll through the list of effects Sounds that appear inthe Sound Browser, and play any one you like. Kyma substitutes yoursample for the one that the effect uses by default. That is anenormous time-saver and lets you experiment with a vast range ofsonic transformations quickly and easily.
TIME IS ON MY SIDE
For anyone doing computer-music composition, sound for film orgames, or even live, interactive performance work, the new Timelineis a major attraction. In fact, it is easily the most significantnew feature in the release. Like a sequencer or multitrack audioeditor, the Timeline offers a two-dimensional, multilayer grid onwhich you place Sounds. But unlike a sequencer or editor, theTimeline lets you build time-varying synthesizers or effects thatcan overlap or morph into one another.
For example, by dragging only a few Sounds into the Timelinefrom the Prototype strip, I created a patch that is a subtractivesynth for 30 seconds, morphs into an additive synthesizer for 60seconds, then becomes a granulating sample player, all with no userinput (see Fig. 5). Because any Kyma Sound can be used onthe Timeline, you can layer processed real-time audio input withsynthetic Sounds and samples. If you want to apply an effect to aSound in the Timeline, just drag the icon representing that effectfrom the Prototype strip or Sound Browser and drop it on top of theSound you want to modify. It doesn't get much easier than that.
There's no limit to the number of tracks or Sounds you can havein the Timeline, though at some point you may run out of real-timeprocessing power (check the FAQ at Symbolic Sound's Web site forperformance benchmarks). If that happens, simply save the entireproject to disk as an audio file (WAV, AIFF, and other common audioformats are supported at rates as high as 24 bit and 100 kHz) oruse Kyma's DiskCache to prerender part of your sound (the synthesislayer, for example). The next time you load that project, theprerendered segment will automatically play back as a sample,greatly reducing the processing demands on the system.
Each track in the Timeline has numerous parameters, whichinclude solo/mute, volume and pan, angle and radius (for surroundmixing), and various routing options. You can create submixes if,for example, you want tracks 2, 4, and 6 to route through a reverbeffect, and it's easy to send a submix or an individual track tothe Capybara's audio outputs. You can also assign any MIDI channelas the source of control data for a track.
The Timeline offers extensive automation features that you canuse to control individual Sounds, tracks, or all events in yourproject simultaneously. At the bottom of the Timeline window aredisplays for graphically editing parameter types. You can draw thecontrol data for a Sound's parameters — for example, areverb's fade time or a granulator's grain size — and you canalso automate any of the track parameters previously mentioned.Also, you can use the BPM parameter to slow down or speed up aproject's playback.
There are numerous tools for manually entering new data andmanipulating preexisting control data, and linking controls iseasy. For example, one parameter might use an inverted version ofanother's control data. You can even import an audio file, haveKyma analyze its frequency or amplitude envelope, and then use thatenvelope to control a Sound's parameters; the entire process takesno more than two or three steps.
In addition to drawing or importing automation data, you can usethe VCS to send parameter values to one or more Sounds as they playin the Timeline, and you can even record VCS-fader movementsdirectly into the automation-editor window. You can perform thesame type of operation using an external MIDI control surface (orsequencer running on the same or another computer), and if you owna CM Labs Motor Mix, you'll find a high level of integration for itwithin Kyma. You can also assign fixed values for each Sound'sparameters before you start playback and simply let yourcompositions evolve unattended.
Kyma has a number of features optimized for surround mixing.These include the ability to set the pan position of both an entiretrack and an individual Sound on a track. Also, you can design asurround mix even if you have only four outputs — just tellKyma to render the Timeline to six or eight surround files on yourdrive. Then send the files to your client, who can load them into,say, a multichannel Pro Tools system and hear the exactspatialization and panning you designed.
Analog I/O (4) XLR balanced standard (expandable to 8) Digital I/0 (2) XLR stereo AES/EBU or (2) RCA stereo S/PDIF Other Connections MIDI In/Out/Thru; (1) BNC word-clock input; (1) BNC house syncinput; (1 pr.) RCA VITC I/O; (1 pr.) RCA LTC I/O Output Level +14.5 dBu Input Clipping Level +14 dBu Resolution 24-bit Internal Processing 24/48/56 (algorithm dependent) Sample Rates All standard rates from 32 to 100 kHz Dynamic Range (A/D)/(D/A) 110 dBA/107 dBA Frequency Response 20 Hz-20 kHz (+0.04/-0.26 dBu @44.1 kHz) Input Impedance 10 k• Crosstalk -110 dB Noise (A/D)/(D/A) 110 dB/105 dB Tuning Resolution 0.0026 Hz Prototype Algorithms >300 Factory Patches >1,000 Dimensions 3U × 16.5" (D) Weight 15 lbs.
Though you can't display video directly in the Timeline, byputting the system in Time Code mode, you can control it from aMIDI program, such as a digital-audio sequencer, or a video deck.Kyma syncs to word clock, LTC, VITC, MTC, and house sync.
NEW MILLENNIUM MODULES
Kyma has always offered numerous ways to generate and processsound, but the new release provides dozens of new Prototype Soundsand a much-improved system for organizing them (see Fig.6). The Sounds are grouped into categories by function andinclude spectral processors, numerous synthesis methods, pitch andformant shifters, and vocoders. A large number of filters,waveshapers, dynamics processors, and looping tools are just someof the list's more traditional algorithms. Keep in mind that anyprocess can be combined with any other — Kyma sets no limitson the type of signal paths you can create or the number ofprocesses you can chain together.
Some of the Prototypes have names that belie their exactfunction. For example, the Spectrum Discombobulator introducescontrollable amounts of random variation into the timing,amplitudes, and frequencies of the sound spectrum, and theFmntOscilBPMFilter KBD is a keyboard-triggered synthesis modulethat generates trancelike, analog-style backgrounds synched to thebpm. The ResynthesizeRandomFrames is handy for freeze-framing yourvocals at the bpm rate, and the SampleBitsBPM happily snips smallbits out of your loops so that you can rearrange them if you sodesire.
For dance, trance, and techno lovers, Kyma includes numeroustempo-based sample players, various sequencer types, and severaldrum machines. The BPMRandoMiniLoops, for example, jumps aroundrandomly within a sample at a rate determined by the bpm settingand picks segments of random durations, playing some forward andsome backward. You'll be the hit of the rave with theDerangedSampleBits Sound, which grabs bits of a loop atuser-defined points and then uses the AnalogSequencer module toplay through those bits in different orderings. Using the VCS (or aMotor Mix), you can toggle steps of the sequence in real time.
Remixing is yet another area that has been enhanced in the newrelease. I set up a patch with three tracks, each containing adifferent loop, then added controls to change the playback tempowithout changing the pitch, the start position within each loop,each track's volume, and the pan position. This set is ready totake on the road, and the production time was a mere 30 minutes.For gathering and preparing source material, there are all sorts ofrhythmic choppers, shufflizers, and live-audio capture tools, whichall work in real time.
OUT OF THE BOX
You'll have no trouble making new Sounds with the Prototypes,but you will certainly spend considerable time with the system's1,000-plus example Sounds. (Also check out contributions from Kymausers at the new users' forum at Symbolic Sound's Web site.) Theexamples incorporate the Prototypes in various ways and areorganized primarily by task. Looking for things that go bump in thenight? Try the Whooshes, Hits, and Bys category. Working on a liveperformance piece? Look into the Tempo, Pitch, and Amp Followingfolder for ways to start and stop a MIDI sequence or audio file,using nothing but your voice. If you really want something unique,let Kyma slow down your voice in real time using its powerful liveanalysis and resynthesis tools. (Hint: look in the Time-scalingcategory.)
There are also examples that show the extensive newspectral-morphing features and more-powerful vocoders, and Kymaeven includes a new set of additive-synthesis examples that showoff the system's power. The Backgrounds and Pads group lookedparticularly intriguing, though I managed to try out only a few ofthe several dozen Sounds in that category. The examples covernearly every Prototype the system offers and are great startingpoints for your explorations.
Kyma's developers have always managed to extract significantprocessing power from the hardware, but this release includes evenmore optimization. That means you get more power from your existingCapybara hardware just by upgrading your software. The Capybaraitself was also significantly enhanced and is available in severalversions.
The Capybara's new 24-bit, 100 kHz converters also soundfantastic, though Windows users still can't gain access to thehardware from their other audio applications. Mac users, on theother hand, can test the beta ASIO drivers available from SymbolicSound. There's currently no support for NT, but users should expectNT and Windows 2000 compatibility later this year.
Kyma's extensive documentation includes a 550-page printedmanual and nearly 100 pages of electronic text relating to the newrelease. The printed manual has some out-of-date information: forexample, it mentions some Prototype Sounds that have new names orcontain renamed parameters. But that is a minor problem that won'tstop you from understanding how the Sounds work. Numerous tutorialsguide you through the system's intricacies, though I wish more ofthem were aimed at the beginner's level.
There's flyby assistance for all the main system controls, andnearly every parameter of every Sound is documented withcontext-sensitive help. Symbolic Sound's Web site also includes afairly thorough FAQ and answers to common questions in the users'forum. Free e-mail and telephone tech support round out thepackage.
COME ON DOWN
The Kyma System is a real engineering feat, and its designersare to be commended for the ongoing and significant enhancementsthey have provided during the past ten years. The system isremarkably stable — I can't recall a single system crash inthe many years I've used it — and when its price is comparedwith that of even a moderately loaded hardware sampler, it lookslike a real bargain.
Kyma stands alone as a mature, hardware-accelerated sound-designplatform. It provides a vast number of tools for nearly any type ofaudio work and the resources for building your own tools when youneed them — no $100 third-party plug-ins required here. Forsound design, live performance, computer-music composition, dancemusic, remixing, and even scientific and engineering applications,Kyma is about as good as it gets.
EM associate editor Dennis Miller lives in the Bostonsuburbs with his wife, two daughters, cat, dog, and sixcomputers.
Minimum System Requirements
MAC: Power Mac; 32 MB RAM; OS 7.05; NuBus or PCIslot
PC: Pentium 120; 32 MB RAM; Windows 95/98/ME
Kyma System5.11 (Mac/Win) sound-design workstation $3,300 basesystem
FEATURES5.0EASE OF USE4.0SOUND QUALITY5.0VALUE4.5 RATING PRODUCTS FROM 1 TO 5
PROS: Vast range of sound-processing and -generatingtools. Intuitive graphical interface. More than 1,000 includedSound examples.
CONS: Insufficient beginner's tutorials. Some Soundsnot intuitively named.