Reading CamillaFIR Output (Current Behavior)
Reading CamillaFIR Output (Current Behavior)
This guide reflects the current code path in CamillaFIR (v4.0.x line).
1. What you get after a run
CamillaFIR produces:
- Interactive dashboards in the UI (Left/Right analysis view).
- A ZIP package saved to
Documents/CamillaFIR/filters/<version>/(safe fallback used if needed):- FIR WAV files (either
L_...wav+R_...wavor a singleStereo_...wav, depending on layout) - Summary report (
Summary_<type>_<fs>Hz.txt) - Convolver configs (
Config_...cfg,camilladsp_...yml)
- FIR WAV files (either
Important current detail:
- ZIP dashboard PNG export is currently disabled by default.
2. Reading the dashboard (UI)
The UI dashboard has 5 panels:
2.1 Magnitude and Alignment
Shows:
- Measured (blue)
- Target (green dashed)
- Predicted (orange)
- Confidence line (magenta)
- Level reference line and smart scan window shading
- Correction range shading (
mag_c_mintomag_c_max)
Read it like this:
- Predicted should track target without narrow high-Q spikes.
- Red-tinted bands indicate low confidence regions.
- Strong LF lift means headroom risk; check Summary headroom section.
2.2 Phase
Shows wrapped phase of the predicted response. Use it mainly to see whether behavior looks smooth around crossover and correction limits.
2.3 Group Delay
Highlights delay structure over frequency.
- Broad LF structures often relate to room modes.
- Sharp local spikes can indicate reflection-related behavior.
2.4 Filter (dB)
This is the FIR gain curve after level-compensated plotting. Use it to confirm boost/cut aggressiveness and low-frequency policy behavior.
2.5 A-FDW Effective BW (oct)
Shows effective adaptive bandwidth when available. If unavailable, the panel displays “No A-FDW BW data (…)”. This is expected in some modes.
3. Reading Summary_...txt
The summary is layered. Main blocks:
3.1 Executive Summary
Includes:
- Acoustic Score (L/R)
- Target Match (percentage + RMS dB)
- Confidence (L/R)
- RT60 wideband (L/R)
- Worst detected event (L/R)
3.2 Core Settings and Analysis Mode
Includes:
- Key settings used for generation
- Analysis mode (
nativeorcomparison) - Comparison grid metadata when comparison mode is active
- Automatic-mode export block when
AUTOwas used:- selected target curve and selection method
- target-selection trial grid (
top-N x trials) - cached or searched best preset summary
- excursion protection seed -> final frequency when auto-tuned
- winner-polish results for
phase_limitandmag_c_min:Phase-limit winner polish: applied (X -> Y Hz, rank A -> B, tested [...] Hz)when the post-search polish improved the winnerPhase-limit winner polish: not appliedwhen the original winner was already optimalMag-c-min winner polish: tested, no change (kept X Hz, tested [...] Hz)when no improvement was found
3.3 Guards and control blocks
Includes:
- Correction Guards (max cut, slope limits, low-bass cut policy)
- Temporal Decay Control (TDC) status and limits
- A-FDW status and effective bandwidth stats
- XO and phase clamp lines
3.4 RT60 and matching blocks
Includes:
- RT60 wideband and optional per-band values
- Confidence and leveling window info
- Target curve match lines for left/right
3.5 Alignment and gain staging
Includes:
- Peak (pre-norm), global offset, auto gain margin, applied auto gain
- Additional sections appended during export:
DSP EFFECTIVE PARAMSLEVELINGHEADROOM MANAGEMENTBOOST/CUT DIAGNOSTICSCLAMP DIAGNOSTICSSTAGE CHECKPOINTSAUTO-ALIGNACOUSTIC EVENTS(when detected)
4. Practical pass/fail checks
Use these first:
Target Matchimproves clearly versus uncorrected.Net boost peak (post global/headroom)is not positive or only slightly positive.hard_clamp/soft_clipcounters are not excessive.- No unexpected large tilt in
LEVELING. Final Max (filter+auto_gain+headroom)stays sensible for your playback headroom.
5. If the result sounds wrong
- Too little bass:
- Check
low_bass_cut_hz - Check effective max boost and blocked boost reason
- Check correction range upper/lower bounds
- Check
- Too bright or sharp:
- Reduce correction range in upper mids/highs
- Increase smoothing / reduce aggressive slope behavior
- Unstable imaging or odd timing:
- Check
AUTO-ALIGNdelay and gain difference - Re-verify measurement consistency
- Check
Use re-measurement with filter enabled as the final validation.
Disclaimer
AI was used to translate this document from Finnish to English.