Omnissiah has two diagnostic aids: the diagnostic cards that explain what the app found in your room, and the Latency Drift Check for tracking down timing problems between audio interfaces. After this chapter you'll know how to read both.
Diagnostic cards
Diagnostic cards are plain-language notes about your room and your measurement: reflections, room nulls, resonances, and measurement-quality problems. They're the app being straight with you. Some room problems can't be fixed with EQ, and it's better to know than to be misled by a "flat" graph. You'll meet them in three places:
- On the results page, after a correction: per-channel notes about anything the correction left untreated, and why. Selecting a channel on the chart filters the list to that channel.
- On the measurement screen (see Chapter 8 — Measuring your system): the same findings appear live below the charts as you measure (or for whichever snapshot you select), phrased as what the microphone is hearing right now. Move a speaker and watch them update.
- During a guided calibration's measuring step, beneath its chart.
Each card shows the channel, an approximate frequency, a short title, an explanation, and often a suggested action. Hover a card to shade its frequency band on the chart above. Their colour shows severity:
- amber: a warning about measurement quality
- yellow (accent): an advisory about the room
- hairline: information (usually "left alone on purpose")
What the findings mean
| Finding | Type | What it's telling you | What to do |
|---|---|---|---|
| Outside the usable range (below/above ~X) | info | The frequency is past your speaker's usable edge (or below/above the crossover), so it's handed to another speaker or left alone. | Nothing — this is expected. |
| Modal / boundary issue | advisory | Room modes or boundary interference (low confidence, no clear dip). | Not an EQ fix — experiment with speaker, sub and listening positions. |
| Peak left uncorrected | advisory | A real resonance the correction chose not to cut. | Add a manual EQ cut there, or re-run correction. |
| Deep null left alone on purpose | advisory | An acoustic cancellation too deep to fill without excessive boost. | Move the speaker, sub or seat; treat the room — EQ won't fix it. |
| Reflection — acoustic, not EQ | advisory | Coherence collapse from a strong reflection. | Angle or move the speaker; add treatment. EQ can't fix a reflection. |
| Possible measurement noise | warning | A narrow spike with a coherence drop — likely a one-off sound during the measurement. | Re-measure that speaker during a quieter moment. |
| Weak measurement on this channel | warning | Low confidence across the band (shows the signal-to-noise ratio). | Re-measure with higher playback level and less background noise. |
| High-frequency limit | info | Air absorption / single-point limit at the very top. | Left uncorrected on purpose. |
The table shows the results-page wording. The live cards on the measurement screen describe the same findings as observations: "Peak at ~250 Hz" rather than "Peak at ~250 Hz — left uncorrected", since no correction has run yet.
The single most useful pattern to recognise: warnings (amber) are about your measurement, so fix them by re-measuring. Advisories (yellow) are about your room, so fix them by moving things or treating the room, not with more EQ.
Coherence and SNR
Two quality numbers underpin the cards and charts:
- Coherence (shown as Confidence): how much of what the mic heard is genuinely the test signal versus noise/reflections, per frequency. High is good.
- SNR (signal-to-noise ratio): how far your measurement rose above the room's noise floor overall. A low SNR is why you'll be asked to raise the level and quieten the room.
Latency Drift Check
This diagnostic checks whether the round-trip measurement latency stays stable between runs. It matters when your microphone and speakers are on separate audio interfaces: their clocks can drift, which stops two separately-taken measurements lining up to a common time reference. That in turn breaks time-alignment between speakers.
You'll find it in the Measurement settings gear (see Chapter 8 — Measuring your system) and in Settings → Audio → Diagnostics.
Running it
- 1Stop any measurement first (the diagnostic drives the measurement itself).
- 2Keep the mic and speaker fixed and the room quiet.
- 3Choose the number of Runs (3 / 5 / 10 / 20; default 5).
- 4Click Baseline, which runs with the audio streams kept open (your measurement noise floor), then Re-open streams, which forces a full stream re-open before each run, exposing the timing shift that separate interfaces introduce.
Reading the result
Each block reports the Latency range (and its spread) and the Datum range across runs, plus the path to a CSV written to the app's diagnostics folder.
- If Re-open streams shows a range much larger than Baseline, drift is confirmed: your input and output are on different clocks.
- The fix is to use one interface for both mic and speakers if you can, or to enable Clock Drift Correction (see Chapter 8 — Measuring your system).
If your mic and speakers already share one interface, you shouldn't see meaningful drift, and you can leave Clock Drift Correction off.
