Regression Verification After Audio-System Changes
Regression verification proves the intended improvement and checks every approved behavior that the change or its interfaces could disturb.
Change Impact Record
Before editing, replacing, moving, updating, or repairing anything, record:
- change identifier, reason, owner, and date
- physical or digital item changed
- inputs, outputs, loads, sources, controls, and vehicle functions it touches
- requirements, limits, presets, evidence, and user instructions that may be affected
- fallback and known-good release
Regression Levels
| Level | Purpose | Example |
|---|---|---|
| Smoke | Confirm correct identity, power, routing, and no immediate fault at protected level. | Channel check after loading a Wondom release. |
| Targeted | Prove the intended change and direct interfaces. | Measure center coefficient, headroom, and output after changing the center matrix. |
| Subsystem | Retest the complete affected source, DSP, amplifier, speaker, or vehicle-function path. | Recheck all front-stage routes after replacing one amplifier. |
| Integrated | Retest cross-system states and functions. | Sources, presets, calls, alerts, sleep, wake, and final-trim response. |
| Full acceptance | Repeat the complete release matrix when architecture, limits, or evidence validity changed broadly. | New source, DSP platform, speaker layout, or major electrical revision. |
Matched Comparisons
A before-and-after comparison is valid only when the conditions needed for the question match. Record:
- reference content and source version
- source and user level state
- preset and processing state
- instrument, calibration, range, and processing settings
- microphone or probe position
- vehicle openings, trim, seats, headrests, electrical state, and temperature
If the conditions differ, label the comparison as exploratory rather than using it as acceptance evidence.
Common Hidden Regressions
- a source-level change moves the downstream limiter or noise floor
- a filter change alters coherent summing headroom
- a replaced amplifier changes gain, polarity, load, noise, and turn-on behavior
- trim removal changes a speaker path, grille clearance, rattle, or water boundary
- a source or phone update changes normalization, volume restoration, reconnect, or call routing
- a successful normal path fails after cold start, sleep, wake, or an alert
Disposition
change proposed
→ impact and regression scope
→ protected smoke check
→ intended-result verification
→ affected regression tests
→ failure correction and retest
→ new approved release recordA regression failure is not a reason to redefine the expected behavior after the fact. Update the requirement only through a deliberate design decision with documented tradeoffs and renewed acceptance.