Dmitry Sches Thorn Review


Without question, Dmitry Sches is one of the most brilliant developers on the planet, so when he drops a new plugin… well, it’s kind of a big deal. Thorn is Dmitry’s second synth, which uses spectral analysis to compute a complex series of harmonics, and that’s just the beginning of the signal chain!

Thorn has a few more tricks up its sleeve, including the Harmonic Filter, which applies a “spectrum curve” to the harmonics in each oscillator, and if that weren’t enough, the Harmonic Filter’s output is sent to either one of two independent analog modeled filters and processed through a series of nine high quality effects, in addition to several other impressive features I’ll discuss in further detail.

Enough pointless preamble. Let’s see what this baby can do…

The Review

Thorn has three Harmonic Oscillators, each having their own Preset Selector and a Harmonic Editor with a Spectrum Table made up of sixteen “frames” capable of generating up to 128 harmonics, and a Command Menu with a long list of options that can be used to edit the Spectrum Table in all sorts of fun ways that are ideal for innovative, experimental sound design. The “Import Single/All Frames” options in the Command Menu will not import the audio content from WAV files but rather the frequency content of the audio, which is summed and then loaded up into one or all sixteen frames. In the bottom left-hand corner of each Harmonic Oscillator, a “POS” (position) knob will morph in between frames to create beautifully complex, evolving sounds that defy classification.

See also: Dmitry Sches Diversion Review

Within the top panel of each Harmonic Oscillator, you’ll find tuning controls for octaves, semitones, up to eight unison voices, a fine tuning control that becomes a unison detuning control when unison voices are active, and a spread control used to pan unison voices apart in the stereo field. Down below the top panel are a number of additional controls including Copy and Paste buttons, a dedicated Sub Oscillator with three optional waveforms, stereo panning and volume controls, independant filter select buttons, a Phase control with three oscillator Phase Modes and a Vibrato depth control linked to the Vibrato LFO.

But wait! There’s more! An “OSCFX” control (next to the Sub Oscillator knob) provides access to eight Spectral Effects (Phaser, Screamer, Wah, Notch, Shift, Comb, Octaves and Primes) and five Real-Time Effects: Frequency Modulation, Ring Modulation, Pulse Width Modulation, Sync and W-Sync, which is similar to Sync, but the synced harmonics are neatly folded up into a composite “window”. So yeah, that’s just one Harmonic Oscillator. Like I said… it’s kind of a big deal.

Next, we have the Noise Oscillator, which is also capable of sample playback! Yes, you can load your own WAV files into the Noise Oscillator via the Preset Selector, in addition to the noise and percussion samples included in the factory library, which is what I love the most about the Noise Oscillator because you can use samples in an FM/RM scenario to modulate Harmonic Oscillator #3. Using complex signals to modulate a simple waveform is a great way to create some beautiful and terrifying sounds. I just wish it was possible to run the Noise Oscillator’s output through the Harmonic Filter, which operates with the harmonics generated within the Harmonic Oscillators instead of real-time audio.

The Harmonic Filter is quite possibly one of the most innovative sound design tools ever implemented in a software instrument. A “spectrum curve” is applied to each of the 128 harmonics produced by the Harmonic Oscillators, and shifted up and down the frequency domain by the “Shape” knob. Also, you can adjust the low and high frequency content with the “Balance” knob. But the real fun begins in the Harmonic Filter Editor, which is similar to the Harmonic Editors in the Harmonic Oscillators, but this time, there is no frame selector and there’s a MUCH higher number of editable harmonics (8192).

The two independant analog modelled multi-mode filters in the Filter Section have sixteen filter types, each having their own distinct personality. There are a few in particular I’ve grown to love, including the Hypercomb filter, optimized for making formant-like sounds. I don’t think I need to explain the controls in the Filter section, which are all very commonplace, such as a “Link” button, Drive, Velocity Tracking etc. By default, the filters operate in parallel mode, but the “F1” button switches to a serial scheme with the output of Filter 1 routed to the input of Filter 2, which is also pretty commonplace. But your ability to modulate parameters like “F1” from within the Modulation Matrix is truly extraordinary.

The Glitch Sequencer is another oddity. Similar multi-effects sequencers are rarely, if ever, included in a software instrument, so this is a very pleasant surprise. Six effects (Sample Rate reduction, Bit Crusher, Repeater, a volume Gate, Low Pass and High Pass filters) are applied to the input signal across sixteen steps, each with their own level faders and a Random button that will arbitrarily scramble effect settings. Other notable features include twenty-two synchronized speeds, synchronization modes, step length, a “Mix” knob adjusting the balance between the dry/wet signals, and a “Pre FX” button that will insert the Glitch Sequencer before the Effects Section when switched on.

The Arpeggiator isn’t breaking new ground, but it has a squeaky clean look and feel that makes for a smooth, unencumbered workflow. The controls are all laid out in a very simple, straightforward design, which also has sixteen steps, each with adjustable level faders similar to those in the Glitch Sequencer, as well as a few more similar controls such as Steps, Speed, etc. The basic controls are all pretty much standard fare (Note, Velocity, Gate, Tie) but one thing that really put a smile on my face is “Poly” mode, which is perfect for gated pads and pulsing rhythms.

I won’t try to catalog every parameter in each of Thorn’s Effects, so I’ll just touch on the most pertinent aspects of each one. Let’s start with the Delay, which has two parallel delay lines, each with a long list of host synchronized delay times. A lot of people have requested the option for unsynchronized delay times, similar to the way the “Sync” buttons in the Flanger, Phaser and Chorus modules synchronize modulation rates or unsynchronize them when switched off.

As you might have guessed, the LoFi module is a bit-crusher and sample-rate reduction processor with a tone control emulating the signal loss in telephone lines. Simple as pie. The Distortion module is a bit more sophisticated, with half a dozen distortion types and a preprocessed Fuzz stage modelled after a classic analog fuzz pedal. Even though the distortion settings in the “Default” patch are a little over the top, I’ve managed to make good use of them when seasoned to taste.

The Reverb has two modes: a Plate algorithm, which is great for FM bells and percussive material, and a Space algorithm, which is splashy and diffuse, and also has additional lowpass and highpass filters in the feedback path. I would have liked some way to curtail low frequencies in the Plate algorithm, but an EQ placed after the reverb is obviously the most effective solution for unwanted frequencies, and if any subtractive EQ surgery cuts into your output, you can boost the resultant signal or apply a touch of gain reduction within the Compressor module, which is surprisingly transparent, as is the Limiter at the very end of the signal chain, which allows you to cushion loud transients produced by heavy compression.

You can arrange effects in any order by simply dragging them into position with your mouse. Similarly, you can drag and drop modulation from any one of Thorn’s envelopes, LFOs or MSEGs onto the target parameter, assign modulation by right-clicking on the target parameter and then choosing a modulation source from within the “Add Modulation” submenu, and further manage your assignments in the 27-slot Modulation Matrix, which might be the most flexible modulation network I’ve ever used. You can even route modulation to each of the sixteen steps in the Glitch Sequencer and the Arpeggiator!

The Verdict

I’m afraid the CPU hit is substantial, with complex patches involving the Harmonic Filter and all three Harmonic Oscillators spiking at fifty percent processing capacity with only four voices of polyphony. But when you take into consideration what Thorn’s spectral engine is doing under the hood, a heavy hit is a compromise I’m willing to make in pursuit of cutting edge sound design, though CPU optimizations are inevitable in future updates, and Dmitry has also been looking into the possibility of a CPU idle mode.

In regard to Spectrum Tables, I personally prefer to create my own with the commands available in the Harmonic Editor instead of using WAV Import, which has a tendency to produce unpredictable results, even though some people might enjoy that unpredictability. That being said, there’s nothing stopping you from rolling up your sleeves and further editing harmonics imported from WAV files.

I wouldn’t say Thorn necessarily fits into the “bread and butter” category, though it is possible to create warm vintage analog sounds in addition to harsh, unabashedly digital modern sounds, so you’ll have no problem with flexibility. Thorn’s strengths aren’t so clearly defined. Even the phrase “spectral synthesis” somehow feels ambiguous. Everything about it cries out for experimental use. Even though it’s quite capable of operating within your comfort zone, Thorn dares you to traverse a road less traveled.

More info: Thorn ($69 until December 31st, $119 regular price)

Dimitry Sches Thorn Review


Thorn's Harmonic Filter is quite possibly one of the most innovative sound design tools ever implemented in a software instrument.

  • Features
  • Workflow
  • Performance
  • Design
  • Sound
  • Pricing
Share this article. ♥️

About Author

Bryan Lake is a sound designer and a musician. He publishes sound design tutorials and sound libraries on his website Sound Author.


  1. Im not sure exactly what specs would be relevant to post but a 6700k on win10 using Bitwig and LR-16 soundcard at 515 samples 44,1kHz Thorn (64 bit version) didn’t strike me as very cpu heavy at all. I would think my patches are very complex with lots of modulation and most modules active. Maybe 5-10 percent if that much but never even close to 50.
    If you compare it to Diversion and Diva, would you say it uses more per average voice or less?

    • I’m using an older system with a puny processor and not much memory, but that will soon change. I said the CPU hit is “substantial”, not system-critical. On an older system like mine, the hit will be very noticeable. Not fatal. But complex patches with any more than 4 voices are going to slow you down.

  2. One my best investment into happyness so far (besides things like Alchemy, Omnisphere 2).

    Atm im running it on an early 2015 MBP and this beauty didnt even hit 10% CPU usage per instance. Pristine and unique audio quialty, actually intuitive UI, really not a single knob which wouldn’t affect the sound characteristics daramatically and most importantly MUSICALLY! Also the available presets are a treasure.

  3. Juan Maria Solare


    Sorry, when you say that ” will not import the audio content from WAV files but rather the frequency content of the audio,” do you mean that I can load in Thorn a sample of Brigitte Bardot saying “Hi” and the oscilator would imitate her voice (saying hi)?

Leave A Reply