| Filename | Latest commit message | Latest commit date |
|---|---|---|
| .. | ||
| ambisonic-aquasynth | ||
| chirplet-sync-decoder | ||
| feedback-calibration | ||
| gpu-fusion-splatting | ||
| live-adaptive-sync | ||
| perfect-machine-hypotheses | ||
| perfect-machine-study-2026-05-23 | ||
| playback-calibration | ||
| sonar-dsp-recursive-refinement-2026-05-25 | ||
| spout-obs-bridge | ||
| visual-spatial-map | ||
| bioacoustic-contestants-2026-05-24.md | ||
| README.md | ||
| sonar-dsp-recursive-refinement-2026-05-25.md | ||
Research Archive
This directory preserves implementation research and hard-won calibration lessons. It is allowed to mention external tools, papers, vendor docs, and failed paths because forgetting those days would be an expensive little act of self-sabotage.
Research is not runtime authority.
Use it to shape native implementation decisions:
- direct camera and audio drivers;
- pinned/native buffers;
- GPU feature extraction, flow, matching, and fusion;
- Fensalir D3D12 compute/render paths;
- Faust/native DSP;
- OBS-facing program outputs.
Audio timing research entry points:
- chirplet-sync-decoder/summary: chirplet transform notes, why dense matched-filter search was cut from the hybrid hot path, and the current dechirp plus FFT/Goertzel decoder shape.
- live-adaptive-sync/summary: passive program-audio synchronization and sample-rate-offset estimation.
- feedback-calibration/summary: online room/speaker/mic response calibration.
- ambisonic-aquasynth/summary: spatial audio and Ambisonic reference material; useful background, not a license to call unsynchronized mics a coherent field.
Do not promote mirrored examples, old probes, or external library demos into the Mimir hot path just because they exist here. If research points at a concept, rebuild the live version in the Mimir/Fensalir/native stack with a clear owner and invariant.