Easy-Effects-Presets

May 6, 2026 · View on GitHub

My Easy Effects input presets.

Presets

  • input/Voice.json — universal voice chain optimized for low WER on ASR/dictation engines (Whisper, Parakeet, Nova, etc.) while still sounding natural on Zoom calls. Chain: RNNoise (full wet, VAD off) → Autogain (target -18 LUFS, 5s history). No VAD-driven gating, so word onsets aren't clipped.
  • input/Voice Calls.json — same denoising + level normalization as Voice, plus a gentle compressor (2.5:1, soft knee, -20 dB threshold, 3 dB makeup) for more consistent loudness on calls. Autogain target slightly hotter (-16 LUFS) so the recipient doesn't have to crank their volume. Chain: RNNoise → Compressor → Autogain.

Legacy

  • input/Speech Processing.json — empty/passthrough; kept for history.
  • input/Speech Processing 2.json — earlier RNNoise + Autogain chain (autogain target -23 LUFS, 15s history, VAD on). Superseded by Voice.json.

Install

Flatpak:

cp input/*.json ~/.var/app/com.github.wwmm.easyeffects/data/easyeffects/input/

Native:

cp input/*.json ~/.config/easyeffects/input/

Then load via Easy Effects → Presets.