Like me, you will be disappointed to find that spectral synthesis doesn’t involve ghosts. I’m telling you this to save you buying Izotope Iris2, in the vain hope of messages from the spirit world. There are none, save your pennies.
Seeing as I have done my money, I may as well review this thing to make money back by the advertisements that infest this blog. What then is spectral synthesis? Have a look at this interface:
No, wait, that doesn’t look at all like my copy. Hang on, try again:
That’s a bit closer. All that black and blue bruise colour scheme confused me for a moment. Now, as you can see, it starts with a sound sample. Instead of looking at the sample in terms of amplitude, how loud the whole is at any moment, the spectral representation considers how much of a particular frequency is there at any moment. If this was as simple as low and high frequency, then you could use familiar high and low pass filters to adjust the frequency domain of the sound – make it ‘duller’ or ‘brighter’. Here you may adjust thin slivers of frequency to carve out areas of the sound, the thinner the frequency, the closer to a distinct pitch you’re defining.
For example if you took white noise, which covers a wide range of frequencies, and carved out everything but a thin band, you get pitched noise.
There’s two main ways of going about spectral design. The legendary Metasynth is about adding signal to sculpt new sound. The spectral editor in Alchemy is better at adding as well – it first converts sound into a ‘painting’ that you can then rework. Iris seems to leave the sound unmodified until you remove some of it. In fact you can’t add signal by painting at all, nor is there resonance.
The brushes are useful, particularly the magic wand and a box that can be made horizontally over a frequency range. But you’re not getting the range of tools that Metasynth offers – no blurs, no sprays, no echoes, just a lot of hard edged boxes and a circle brush. And because the program doesn’t think about the sample as an image, you can’t send the image to Photoshop to edit it there. It’s not unreasonable given the way the sound quality is kept clean, but you’ll need to know that when deciding which to use.
Since version 1 iZotope have made two big changes. They’ve included all the sample libraries that were sold with the previous version. As much as that’s pissed off early adopters, it’s good for people jumping aboard now. On first inspection these libraries are OK but not great – lots of rumble, lots of too quiet sounds – I’m tempted to batch the lot through Sound Forge to make them work better.
More importantly they’ve added a modulation system that is a direct lift from NI’s Massive. You have 5 envelopes for example, just the right number for four sample layers and the main volume. The LFO’s are nice, you can see that they are modelled on analogue signals and they’re stackable to get complex variations. I was pleased that they control the effect units as well (weird echoes ahoy). This livens up the sample playback no end.
A concern with drawing the filtration directly on samples is that durations are key tracked, leading to chipmunks. How iZotrope have dealt with this is unpopular in the message boards, as it’s not a full time stretch, instead something called Radius RT that stretches the duration without compensating the frequency domain. I personally think it obvious that you don’t want to shift frequencies around at all if you’re trying to pick them out of a sound. You can use a traditional filter on the result, although there’s only one for all notes combined.
All up it’s a decent way to get a certain kind of sound; inharmonic, sharp, shimmery. It is a case where software makes sense, and despite a whole bunch of included analogue hardware samples, not really a good tool for warm and phat, which Massive already does thanks very much. Alchemy is still the better purchase for creating from scratch. I give Iris 3 out 5 Nuke controllers.