OTT Disting NT Plugin

Demo: https://www.youtube.com/watch?v=6zNhZXQEgkU

Demo jam: https://soundcloud.com/tsarpf/disting-nt-ott-style-compressor-test

This project builds a native C++ OTT-style multiband compressor plug-in for the disting NT. The DSP implementation lives in ott_algo.cpp and ott_dsp.h, with the custom interface in ott_ui.cpp. It uses CMSIS-DSP for the crossover filters.

Using the plug-in

Make sure you have disting NT firmware 1.15 or newer, then copy ott.o from this repository to /programs/plug-ins/ott.o on the disting SD card. It will appear in the algorithm list after the plug-in is scanned and loaded.

Building it yourself

Clone this repository with the disting NT API submodule enabled:

git clone --recursive <repo-url>

If you have already cloned without --recursive, initialise the submodule with:

git submodule update --init --recursive

Building

The Makefile expects:

• the Arm GNU toolchain (arm-none-eabi-g++ and arm-none-eabi-gcc) in PATH
• CMSIS-DSP in CMSIS-DSP/
• CMSIS Core in CMSIS_6/

The CMSIS locations can be overridden with the CMSIS_DSP and CMSIS_CORE make variables. The versions used for this build can be fetched with:

git clone --depth 1 --branch v1.17.1 \
  https://github.com/ARM-software/CMSIS-DSP.git CMSIS-DSP
git clone --depth 1 --branch v6.3.0 \
  https://github.com/ARM-software/CMSIS_6.git CMSIS_6

Build and run the native host tests with:

make clean && make
make test

This builds ott.o, which can be loaded on a disting NT device.

Environment and limitations

This plug-in targets the disting NT hardware. The Makefile builds ott.o using partial linking (-r) and disables exceptions and RTTI. All per-instance state is stored directly in the SRAM block provided by the disting NT host; the plug-in does not install a custom allocator or depend on a C++ heap. See distingnt_api/include/distingnt/api.h for the full API.

Xfer OTT reference renders

tools/ott_reference.py renders deterministic transfer curves, crossover impulses, timing steps and control probes through Xfer OTT without opening its editor or an audio device. It requires NumPy, DawDreamer and a native host matching the plug-in binary (for example, Windows Python for OTT_x64.dll):

python tools/ott_reference.py /path/to/OTT_x64.dll \
  --output /tmp/xfer_ott_reference

The command writes numerical metadata to JSON and the source/reference audio to a compressed NPZ file. Add --extended for the slower branch-isolation, band-makeup, crossover and stereo-link probes used for detailed calibration.

OTT dynamic reference validation

The OTT detector and two-sample lookahead are fitted to isolated Xfer dynamics. Factory Attack is 10.0 ms in all bands; Release is 64.8 ms in low/mid and 30.3 ms in high. Existing presets retain saved values, so use these times or create a fresh instance to get the measured default sound.

make test includes measured transient-envelope and peak regressions. For a larger offline comparison, build make build/ott_render, then use tools/ott_dynamics.py capture, compare, and export-tests. See the measurement and correction report for commands, results, and remaining differences.

Ableton Live reference renders

Live Probe runs batches through native Live effects or hosted plug-ins using two generated Max for Live devices. The Python runner plays generated probes or existing WAVs, sweeps parameters, captures stereo float WAVs with parameter readbacks, and restores the initial settings. It includes a Vocoder Modulator-mode width/depth sweep and a generic effect example.

The native vocoder uses a measured fourth-order filterbank and sustained envelope contrast. Its before/after report compares the current implementation with Live and the previous DSP.

License

Original code in this repository is MIT licensed. Third-party dependencies keep their respective licenses.