spectral Machine delay


Once you’ve installed and instantiated, Spectral Machine will present you with a simple dialog box. On the left are two checkboxes that divide the effects into General Purpose and Monophonic categories. This is an either/or proposition, and there are different specific effects that appear depending on the category you select. Note that the Monophonic button doesn’t refer to monaural, as in one channel. It means one sound, like a single voiceover talent, or a clarinet, or a guitar playing a solo melody. A mono file will work best as well, but the recording itself needs to be of a single voice. On the other hand, the General Purpose presets work better with other stuff containing multiple “voices,” like a strummed guitar or two people speaking or singing together. Most music you’ll want to process will come out much better using the General Purpose preset, while your voice talent will give better results using Monophonic.

Below those two buttons are the effects themselves, and for the most part the name gives a good idea of what’s about to take place. Highlighting any one of these will change the controls that live in the center of the interface. Different presets will show different numbers and types of controls, but most consist of faders or knobs for adjusting the various parameters. Incidentally, any parameter that has a numeric display can be double-clicked and the desired value can be typed in. There’s also a handy description of each preset and its associated controls that appears along the bottom of the interface. In many cases the description also has tips on what parameters to change for the best results. Finally, on the right are two large faders for effect mix (wet or dry) and for output gain. There’s also a clip light above the faders, although I only saw that light up once during my evaluation period. There’s no indication as to what level causes the clip light to go off, so I would assume it’s set to 0 dBFS within the plug’s structure.

You’ll probably want to use Spectral Machine on a channel insert, although you could put in on an aux track that was fed from sends on various channels. The one thing to watch for is clipping on the insert or bus; some of these presets can create really nasty sounds, particularly at the extremes of some parameter settings. That’s where the gain fader on the plug comes in handy, although it’s worthwhile to check the input level to the plug as well. Most inserts in editors can be used either pre-fader or post-fader; to guard against clipping the input of the plug, it’s wise to make sure you’re using it post-fader which is the default for the majority of editors. So what does it do, and how does it sound? Below are some of my own descriptions of settings I like a lot.


One of the signature effects of Spectral Machine would have to be Delay Spectral Bands. Here you set the low- and high-cutoff frequencies, ideally leaving some space in the midrange section. You then set a delay time and regeneration (aka feedback, which controls the number of repeats) for the delays, to be applied to each frequency band. Setting the delay times to different values results in echoes of the frequency bands that are each clearly audible. This is a wicked-good effect for the end of a fast, hard-hitting promo -- my personal fave was to give the mid and high enough feedback that they hung out over the end of the spot, creating a couple of phone-filter sounding repeats of the last bit of a mixed promo. This effect can get messy-sounding, especially if the low-frequencies are repeated too densely, but the mids and highs work really well. On the other hand, a promo that ends with a nice low boom sounded much thicker with low-frequency delay audible. Note that these parameters are automatable in most editors, so you can bring the effects in and out in precise ways.

Another stunning preset in this category is the Sample & Hold. Like an old-school synthesizer, this preset stores a snapshot of the sound at a particular moment in time and loops it into a tone, which decays over time and is nearly inaudible when the next snapshot is taken and the process begins again. The only control you have is the hold time, which sounds to me to actually be the time between snapshots which is measured in milliseconds; the max time is 1000ms (one second). The overall effect is rhythmic and robotic at the same time, and works really well with the Wet/Dry control set to between .5 and .75 (full wet is 1.0). My only complaint about this preset is that there’s no way to sync it to a music bed except by trial and error. It would be great if it could be set to beats per minute for effecting music whose tempo you knew or could figure. Perhaps one could even set it to the master tempo of the session in the editor, but BPM would be good enough for me.

The last one I’ll mention here is the Oscillating Peak/Notch, which can be set up to sound like a frequency-dependent flanger, and represents another nice one with complex material. I fed both noise and music into Spectral Machine with this preset and recorded the output. A snip here and a trim there, and I’d created some really interesting sweeps and whooshes in nothing flat. I can see this being an effective way to renew some tired and overused effects with little effort.


Trust me, you do not want to run music through these presets unless you’re a fan of artifacts -- you’ll get a lifetime supply with a music source. On the other hand, most voices sounded acceptable-to-great with these presets relative to artifacts. Look, it’s not supposed to be AutoTune, ya know?

Standouts in this category would include Harmonize, which does a musically-acceptable job of creating up to two artificial harmonies with a voiceover track. However, if you ignore musical harmonies and instead set it up as a frequency-specific doubling effect, using harmonies that are less than one semitone above or below the original voice pitch, then delay these relative to the original, you get a really nice doubling effect that is in some ways more convincing that most time-based doubling effects. Keep the delay time short, and use the Wet/Dry control to determine how “doubled” you want the track to be, and you’ll get a fatter track that maintains good articulation so the VO can be understood.

On the other hand, the Spectral Shape Shifter is just weird. The description says it changes the formant filtering in the voice without affecting the pitch, like making a big man’s voice sound like it’s coming out of a small woman’s mouth without changing the voice pitch. I guess I’ll buy that, although there’s a fairly narrow range of settings that produce something that is recognizable as human speech. It’s not that there are too many artifacts; this preset really does mangle the sound in some interesting ways. Don’t expect this one to work as described -- just play with the settings and you’ll find some strange noises that cannot be created any other way as far as I know.

The documentation consists of a 45-page PDF, which is both clear and complete. There’s not a lot of theory here, but in terms of helping you get some useful noises out of Spectral Machine it does the job well.


Other than the woolly or crashy character of some of the presets (especially the Monophonic ones), which is to be expected in a plug as wild as Spectral Machine, I do wish there was some way to monitor the input and output levels. An “LED” on each end would be useful; a meter of some kind would be a delight. But Spectral Machine is definitely a trial-and-error business anyway, so one could postulate that the input and output levels should be adjusted to see what happens in any case.

If you’re looking for something completely different-sounding than the stock effects in your editor, or the same Waves-bundle effects that everyone else uses, give Spectral Machine a shot. Download the demo and put some tracks through it. As far as I’m concerned, it represents a good use of $75 in entertainment value alone. You will get some interesting and/or useful noises from it, I promise. By the way, if you decide to buy it, the company offers a healthy discount on its other plug-ins if you buy them at the same time. Steve sez check it out.

Spectral Machine is available from the company’s website at www.sfxmachine.com, and carries a retail price of $75 USD.