Kantos is a plug-in with a difference its synth engine is not triggered via MIDI, but by an audio input, from which it also derives its pitch. Is it a uniquely creative use of Antares' pitch-tracking technology, or merely a novelty one-trick pony?
Antares' first venture into virtual instruments, Kantos (which comes from the ancient Greek word meaning 'to manually propel an empty metal soft drinks container') rather unsurprisingly incorporates the company's proven pitch-tracking technology, as seen in their now-ubiquitous Auto-Tune software and hardware products. Indeed pitch-tracking is the only way to play Kantos rather than designing the system to be controlled via MIDI in the usual plug-in fashion, all the necessary pitch and trigger information is derived from an audio input.
Kantos runs under VST, MAS or RTAS on compatible Mac or PC platforms and is authorised using the now familiar challenge-and-response system. The program will run for 20 days prior to authorisation, so you can use it as soon as you install it. Like most software instruments, Kantos uses a fair amount of CPU power; a computer with a fairly fast processor is essential if you want to use multiple plug-ins at the same time.
Despite a graphical user interface that looks like a sneeze in zero gravity, Kantos is based on a fairly straightforward two-oscillator synthesizer topography where each oscillator can generate a waveform selected from a wavetable menu plus a further fundamental sine wave. There's also a single noise source. A mixer section allows the levels of the oscillators, the noise and the original audio to be combined along with the onboard delay effects. Conventional filters and envelope shapers are used to modify the oscillator sounds, and there's a very easy-to-use modulation matrix that allows any one of eight sources to be mapped to a choice of destinations with variable modulation depth. Where Kantos differs from conventional synthesizers, apart from the use of an audio trigger source, is that the timbral characteristics of the source are monitored and may be used to 'articulate' the oscillator and noise sounds in way that is both reminiscent of, and entirely dissimilar to, a vocoder. You have to hear it to understand what I mean by this!
The audio used to trigger the synth must be monophonic and of a reasonably uniform loudness if accurate pitch-tracking is required, though some interesting atonal chaos can be had by feeding drum loops or polyphonic sounds into the instrument. Normally an audio track will be used as the triggering source, in which case it should be as free from background noise and crosstalk as possible and preferably normalised. Guitar works well as a trigger source, but you have to take care not to allow notes to ring on when you pick a new note, as this confuses the software. I also tried Kantos with a live audio signal, but found this introduced a noticeable tracking delay causing each new note to jump up to the correct pitch shortly after triggering. Presumably this doesn't happen when working from audio tracks, as Kantos gets a preview of the audio moments before having to output the pitch.
Antares Kantos £266
Pitch-tracking synthesis with articulation.
Huge range of sound possibilities.
Rather dark user interface could make extended use hard on the eyes.
Sound source and input settings have to be selected with care.
Kantos offers something truly new in plug-in synthesis and the little effort that's needed to get good results pays off with sounds that can't be achieved by any other means.
By default, Kantos tracks all the pitch variations and inflections of the source sound, but if you want something a little more controlled, there's a miniature one-octave keyboard for each oscillator where you can enter the 'legal' notes for the tune, much as you can with Auto-Tune. Using this function strips away the pitch inflections of the original sound and turning on all the keys produces a chromatic Quantise effect. The keyboard only defines notes, not octaves, and with some sound sources where the fundamental pitch is not well-defined, Kantos can pick the wrong octave. I noticed this using some female vocal tracks as sources, but in most instances, voice works well as a source, with the added advantage that the articulation feature allows some speech intelligibility to be imposed upon the synthesized waveform.
Articulation, Effects & Envelopes
Articulation is applied in a technically very complex but operationally very simple way. A graphical window functions as a kind of virtual joystick which sets the degree and character of the formant processing, so all you have to do is move the ball around the display until you get a sound you like. When the noise signal is passed through the articulator, the result is a weird robotic whispering effect, which works well added to the oscillator sounds, especially when vocals are being used as the source.
No synth would be complete without LFOs and Kantos includes two multi-waveform types that can be routed via the modulation matrix (shown above). Their frequencies can be adjusted using sliders, or alternatively, there's a tap-tempo function.
The dynamic envelope of the synth sound is normally controlled by the level of the input signal, but again there are two ADSR envelope-shapers that can be routed via the modulation matrix to control a choice of functions. For ADSR level control, an on-screen On button forces the ADSR to control level in addition to whatever other function it may be routed to in the matrix. Both oscillators have independent octave and semitone pitch-switching as well as a Glide function, so pretty much all the traditional synth parameters are available.
Effects can gobble up CPU capacity, so Antares have wisely kept away from reverb and instead stuck to offering delay and chorus, both of which are relatively economical in processing terms. A separate chorus effect is available for each oscillator while the delay effect is common to both. Delay time can be set in beats per minute, locked to the song tempo or entered as a value in milliseconds.
Because Kantos functions using an audio source, it is deployed as an insert plug-in rather than as a virtual instrument. My first impression was that the user interface was a case of 'design before functionality', and even after getting used to it, I still find it unnecessarily dark. However, the signal flow is fairly easy to follow and the little blobs of light on the green 'sneeze' lines function as fader handles making adjustment easy. While setting up, it helps to use one oscillator only and to pick a fairly simple synth waveform, such as the default sawtooth, or a square wave, to help optimise the detector thresholds. I also found that when using guitar as a trigger source, the tracking was better when using a bright sound, so if you experience difficulty with any source, inserting an EQ before Kantos may help.
While clean sounds are tracked surprisingly accurately, spurious noises are not treated so kindly, so if you are using a guitar track as a source, it may help to use your sequencer's waveform editor to silence any excessive fingering or pick noises that occur between notes. I also found it best not to use the little keyboard for scale quantising when using instruments like guitar or voice where large amounts of pitch-bend are being used, as these often translate into trills or, in chromatic mode, trills to the wrong note. If the source needs gentle pitch-correction, then inserting an Auto-Tune plug-in before Kantos works well, as Auto-Tune gives you much more control over the rate of pitch correction.
During my tests, I tried Kantos with male and female voices, guitar and wooden flute as sources, and aside from detecting the wrong octave for some of the female vocal notes, it performed pretty well. I was particularly impressed by the way guitar and flute could be turned into quite realistic-sounding bowed instruments the articulation feature definitely helps here but don't make the mistake of thinking you can throw any old audio at Kantos and then get great results without making some adjustments. The input detector thresholds are critical to good performance, as is the use of clean, monophonic material, but as stated earlier, the chaotic results obtained when feeding in drum loops can be quite interesting. I also liked some of the effects that could be achieved by layering an untreated guitar sound with the synthesized output for an unusual double-tracking effect. Even the basic synth waveforms produce plenty of variety, so experimenting with the couple of dozen or so included wavetables widens the scope enormously, and if you get bored of those, there are more on the installation CD, and more still on the Antares web site.
No device that follows audio pitch is ever going to be a plug-and-go solution, but Kantos works well on the majority of pitched sound sources, providing they are strictly monophonic and include no delay-based effects. You will have to work at the input settings and may have to edit your audio tracks to remove spurious noises and sounds, but the reward is a unique type of synthesis that follows many of the nuances of the original signal. I'm still not convinced by the user-interface design, but once your eyes have adjusted to the 'snot-in-outer-space' styling, it's not difficult to master.
To date, Kantos is unique, so if you want to experiment with synthesis that follows both the pitch and articulation of an audio track, this is the only way to go. Sometimes the results are hit and miss, but if you select your source sound with care and are prepared to experiment with EQ and perhaps compression, you can get excellent results in a short space of time. You'll soon learn what works and what doesn't, and unlike conventional synthesis, the resulting sound has an organic quality that I've not heard from any other instrument. To sum it up, Kantos is pretty weird, but in a very creative way.
800MHz Apple Mac G4 with 768MB of RAM running Mac OS v9.2.
Emagic Logic Audio Platinum v5.1.3.