CamillaFIR Guide (Recommended AUTO Workflow)

This guide reflects the current CamillaFIR workflow in v3.6.5: measure in REW, keep AUTO mode, let CamillaFIR search for a good preset, export FIR filters, and verify with a new measurement.

Quick workflow

Measure Left and Right in REW.
Export REW data as TXT with magnitude + phase, or use the WAV/IR workflow.
Open CamillaFIR and keep AUTO mode.
Leave the filter type at Asymmetric unless you have a specific reason to use another type.
Choose an AUTO goal (balanced for most rooms, subwoofers for sub-only work), then set optional target and HPF preferences.
Press START, review the winning preset, and export the ZIP package.
Load the WAV filters into your convolution engine.
Re-measure and validate.

1. Prepare measurements in REW

Load the correct microphone calibration file in REW before measuring.
Measure Left and Right channels separately.
Keep measurement procedure and timing reference consistent between channels.
Avoid clipping and bad SNR.
For TXT export, include Frequency + Magnitude + Phase.
If you use the WAV/IR import workflow, use REW export settings: Mono, float32, Normalise, Place t=0 (256).
Header/comment lines are supported (*, #, ; are ignored).

Tip: good input data matters more than aggressive correction.

2. Import into CamillaFIR and keep AUTO mode

CamillaFIR has three operating modes:

AUTO: recommended for most users; automatic target selection and preset search.
BASIC: guarded manual workflow.
ADVANCED: expert manual workflow with fewer policy limits.

Why AUTO is the recommended starting point:

It can auto-select a suitable built-in target curve if you do not lock in your own target.
It searches multiple candidate presets and applies the best-ranked winner before export.
It reuses cache hits when the same measurements and relevant settings are seen again.
It exports summary metadata about the winning preset, target choice, and run details.

Important:

Selecting a mode alone does not rewrite all visible values.
Use Apply mode defaults only when you want to reset the current UI values to that mode’s baseline.
Fresh configs start in AUTO.
Fresh configs also start with Asymmetric as the default filter type, which is a good general-purpose choice.

Current AUTO goals:

balanced: recommended default for most rooms.
room-safe: more conservative in difficult rooms.
low-ripple: prioritizes smoother bass behavior around room modes.
flat: prioritizes the flattest measured result.
subwoofers: subwoofer-focused AUTO goal that forces Smart Scan leveling to 20-200 Hz.

3. Set the search constraints, not every value manually

In AUTO, the visible UI values act as the search baseline and constraints. You usually do not need to tune every field by hand before running.

Practical starting point:

Filter type: Asymmetric for most systems.
Goal: balanced unless you are doing sub-only work or solving a clearly difficult room.
Max boost: keep it conservative, usually +3 dB to +5 dB.
Correction range: keep correction mostly in bass/lower mids unless your measurement quality is very high.
Target curve: let AUTO choose from built-in targets, or explicitly select your own preferred target.
Max boost: In AUTO the visible max-boost control does not directly determine the applied boost; AUTO enforces its own safe boost limits during the search and aims to keep boosts conservative.
Correction range: keep correction mostly in bass/lower mids unless your measurement quality is very high.
Target curve: let AUTO choose from built-in targets, or explicitly select your own preferred target.
HPF: optional in AUTO. If you enable HPF in AUTO, the program will select optimal HPF frequency and slope automatically; the UI enable/disable flag is respected but frequency/slope are determined by AUTO.

Filter type guidance:

Asymmetric: recommended default and best general balance of correction quality, stability, and latency.
Linear Phase: use if maximum linear-phase behavior matters more than latency.
Minimum Phase: use for lower-latency causal correction.
Mixed Phase: use if you specifically want mixed-phase behavior.

4. Use safety features on purpose

CamillaFIR still relies on safety limits even when AUTO is doing the preset search.

Key protections:

Max boost/cut limits
Global effective boost cap (8 dB maximum safe boost)
Slope limits
Excursion protection
Low-bass cut
Optional HPF
TDC (Temporal Decay Control)
A-FDW
Stereo link

During automatic runs, CamillaFIR may manage or lock some controls to keep the search valid and safe.

Do not try to fix deep nulls with heavy boost. Placement, crossover work, or room treatment is usually more effective.

5. Run AUTO, then export filters

Press START to begin the automatic workflow.

What happens in practice:

CamillaFIR may evaluate built-in target curves first if you have not locked a target.
It runs automatic preset search and refinement passes.
It applies the best-ranked winner to the final result view and export bundle.
Repeated runs with the same measurements and key settings may reuse cache data and finish faster.

Export creates a ZIP package in the default export folder:

Documents/CamillaFIR/filters/<version>/

If that location is not writable, CamillaFIR uses a safe fallback path and shows the final path in the Results view.

Typical package contents:

L/R FIR WAV files (32-bit float)
Summary_...txt report with AUTO-mode metadata
CamillaDSP config snippet (.cfg)
CamillaDSP .yml (single-rate export, or multi-rate variant)
Dashboard PNG files (or TXT fallback if PNG is unavailable)

Multi-rate export targets:

44.1 / 48 / 88.2 / 96 / 176.4 / 192 kHz

6. Load filters into your DSP

CamillaDSP: load L/R WAV files into the convolution pipeline, or use the generated YAML.
Roon: load FIR WAV in Convolution settings.
Equalizer APO: use the Convolution module and keep enough preamp/headroom.

7. Verify after applying filters

Always re-measure with the filter active:

Confirm target match improved.
Check headroom and clipping risk.
Review Summary.txt for winner details, confidence, decay, and clamp diagnostics.
Adjust goal, correction range, or boost limit if the result sounds or measures over-corrected.

Common mistakes to avoid

Switching to BASIC by habit when AUTO would be the better starting point.
Overboosting bass/nulls.
Correcting too wide a band with low-confidence data.
Ignoring latency implications of filter type.
Skipping post-filter verification measurement.

docs/Official_Manual.md
docs/Modes.md
docs/CamillaFIR_Reading_Output_Guide.md

Download

https://github.com/VilhoValittu/CamillaFIR/releases

Disclaimer

AI was used to translate this document from Finnish to English.