The cheapest credible portable synthesizer on the market is the Korg Volca series, around $150 USD, with a single synthesis engine and a 16-step sequencer. The serious portable rigs (Elektron Digitakt, Polyend Tracker, the OP-1) start at $700 and go up. The cheap end is a toy. The serious end is a luxury. There is no $50 instrument that runs three synthesis engines, samples at high resolution, and survives a full day in a park on one battery.

That is the design problem.

The SMPL Synth is an attempt to occupy that empty slot. Three synthesis engines (FM, additive, subtractive), 1 GB of sampling memory, 40+ hours of continuous battery life, a two-octave keyboard, four rotary encoders, and an outdoor-readable 400×240 monochrome display. Target retail price: $50 USD.

The marketing copy makes that sound like a magic trick. The interesting question is not whether it works (it does, in a partial prototype). The interesting question is which architectural calls let you get there.

The constraints

Cost ceiling: $50 retail means roughly $20 bill of materials at scale. Battery: 40 hours of continuous use means an average current draw of about 30 mA on a single 18650 lithium cell, with synthesis, audio output, and display all live. Display: sunlight-readable monochrome with effectively zero static draw. Audio path: codec quality competitive with the $300 reference, not with the $20 toy.

Each of those constraints kills the obvious solution for the others. A bright color display blows the battery target. A discount codec kills the audio claim. Heavy DSP on the MCU blows the cost target by forcing a higher-tier part. The whole project lives in the narrow intersection of good enough to be musical and cheap enough to be irresponsible to a hobby budget.

The architectural calls

Codec. Audio quality on a portable synth lives entirely in the codec. The SMPL uses an Analog Devices part from the consumer-audio family, the class you find in the $300 portable recorders, not the $20 toys. This was the single most expensive component decision and the one that determined whether the project was a serious instrument or a novelty. Everything else in the audio path follows from this choice: the I²S bus, the anti-alias filter topology, the analog output stage.

Display. A sunlight-readable monochrome panel (Sharp Memory LCD class) holds its image with sub-microamp static current and refreshes only when content changes. Compared to a backlit color OLED, the power savings are roughly two orders of magnitude. The 400×240 resolution fits everything an embedded synth's UI actually needs: waveform views, a modulation matrix, the keyboard overlay, parameter readouts. Color is a nice-to-have on a desktop screen and a battery killer on a portable.

MCU class. Running FM, additive, and subtractive synthesis simultaneously at musically usable sample rates requires a Cortex-M7 with floating-point and DSP instructions, in the 400–600 MHz class. Anything slower forces compromises (lower polyphony, fixed-point synthesis math, no oversampling). Anything faster blows the power and cost budgets. The SMPL targets the middle of this range. The chip choice is the one that, in retrospect, dominates every later decision: it sets the floating-point headroom, the DMA-capable I/O, and the deep-sleep current that lets the 40-hour battery target survive real use.

Memory split. 1 GB of internal flash for samples is generous on a portable; the working set in SRAM is small and tightly managed. The codec talks to the MCU over I²S; samples stream from flash via DMA; the synthesis kernels run from SRAM with double-buffered output. The bus topology is the kind of detail that doesn't show up in the marketing copy but determines whether the device clicks audibly between voices or doesn't.

Three synthesis engines. FM is computationally cheap, a small set of operators with phase-modulated oscillators. Additive is expensive in the naive form (sum of many sinusoids) and reasonable if you back it with a wavetable approximation. Subtractive is moderate, a couple of oscillators feeding a biquad or ladder filter. Each engine fits a different musical aesthetic and each costs a different fraction of the CPU budget. Choosing all three means budgeting all three; the tradeoff appears as polyphony per engine, which is exposed in the UI.

Status, honestly

The hardware is in active prototyping. The marketing site is live with pre-order signups. There is a working firmware that runs FM voices through the codec; additive and subtractive are scaffolded but not finished. The 40-hour battery target has been measured on a benchtop; field testing is in progress. The $50 retail target is contingent on production volume the project has not yet committed to. None of that is hidden on the site at smpl.ma-r-s.com. The pre-order form collects reserved orders, not charged orders.

The argument

SMPL is a project about the gap between what portable synth hardware costs and what it could cost if you made it from first principles instead of inheriting design debt. The codec is the cheap part. The expensive part is everything else most manufacturers don't think to question. Do you actually need color? Backlight? GPU? Capacitive touch? A phone-grade SoC?

The answer is no, and the answer has been no since the 1980s, and the products at the cheap end of the synth market still pretend it's yes.

The instrument is what the project produces. The argument is what the project is. Source on GitHub; live site at smpl.ma-r-s.com.