Wild Thing on the Test Bench

Method

• The installed native Windows VST3 binary is pinned by SHA-256 and matched to its installation receipt. Actual loaded processor name and module path are checked in a separate REAPER process.
• REAPER runs on an isolated non-input Windows desktop, using its own resource folder. No FX editor, audio-device playback or existing REAPER process is used.
• Parameters below are the actual defaults read back after loading. Reported latency and host-program count are read directly through the REAPER API.
• Original calibration: one second silence, four-second logarithmic sine sweep from 20 Hz to 20 kHz at -24 dBFS with 10 ms edge fades, one second silence, two seconds 1 kHz sine at -18 dBFS, then one second silence. Input is stereo 16-bit PCM at 48 kHz.
• Processed and host-bypassed projects are saved separately and rendered offline to stereo WAV at 48 kHz. The whole recordings are decoded and checked for length, finite samples and non-silence.
• Sweep magnitude: FFT output/input ratio, using the four-second sweep section, 262,144-point FFT and median magnitude in a 1/24-octave band around each reported frequency. No response smoothing outside those bands.

Limits of this report

• Sweep transfer includes the behaviour of dynamic and nonlinear processing at this signal level. It is not a universal linear frequency-response specification.
• The 16-bit calibration input has 0.010627% measured THD. Output THD includes stimulus distortion; no claim is made about plug-in-added distortion below this floor, audibility, other levels or other products.
• No true-peak ceiling-overshoot result has been measured. The calibration does not stress a limiter ceiling.
• Real-time CPU at a 128-sample buffer has not been measured. The isolated offline route uses a 512-sample device configuration and does not start playback; render speed is not reported as CPU usage.
• Windows VST3 only. This report does not measure the macOS build or imply analogue-hardware equivalence or an endorsement.
• Signal measurements are not subjective listening. The remaining measurements are listed here rather than represented as completed.