Why acoustic tests need a method
Listening is the final judge of an audio product, but listening alone cannot hold a production line to a standard. Two engineers in two rooms with two playlists will disagree, and a lot that sounds slightly different from the golden sample is hard to reject on impressions. Acoustic tests solve this by measuring the same things, the same way, every time. Their value depends entirely on the method: a figure measured on one fixture at one distance cannot be compared with a figure measured elsewhere. The first deliverable of any acoustic test plan is therefore a written method, agreed between the buyer, the desk and the partner plant before samples are compared.
A method names the fixture or head simulator, the microphone positions, the test signals, the volume steps, the room conditions, the firmware and tuning file version, and the pass rule. For batch inspection the pass rule is usually a tolerance band around the golden sample rather than an absolute figure, because the aim is consistency with what was approved.
The checks and what they tell you
| Check | What it measures | What a problem looks like |
|---|---|---|
| Frequency response | Output across the audible range at the ear position | Thin low end, harsh peaks, mismatch between temples |
| Maximum output | Level at the maximum volume step, per agreed method | Too quiet, or above the agreed ceiling |
| Distortion | How much the driver adds unwanted harmonics | Crackle or harshness on loud passages |
| Rub and buzz | Mechanical faults in the driver or temple | Rattles, buzzing at certain tones |
| Sound leak | Level at a set distance from the wearer position | More leak than the golden sample |
| Microphone sensitivity | Level and response of each microphone | One channel weak, blocked or unsealed port |
| Call noise reduction | Voice clarity with background noise, per firmware | Noise passes through, voice sounds processed |
| Echo | Far-end voice returning through the microphones | The other party hears themselves |
Speaker frequency response, leak and microphone noise reduction are standard items in the category test plan; the others are added per project. For the design background on leak and output, see open-ear speaker design and audio glasses volume and SPL.
Fixtures and conditions
Glasses are harder to measure than earbuds because the output depends on where the temple sits relative to the ear. Plants typically use a head and torso simulator or a simpler head fixture with ear-position microphones, plus a fixed mounting so the glasses sit the same way every time. Leak is measured with an additional microphone at an agreed distance and angle, often in a quiet room or small test box. Microphone checks use a mouth simulator or a reference loudspeaker at the mouth position, with background noise played from fixed speakers for noise reduction tests.
Conditions to write down include room noise, temperature, which frame size was tested, battery charge level, and whether the test ran over Bluetooth from a reference source or through a wired test mode on the board. Production lines often use a test mode for speed; it should be correlated once with Bluetooth playback so the two agree.
Tests by sample stage
Acoustic checks are not run with the same depth at every stage. Engineering samples get the full set to choose a design; the golden sample gets the full set to fix the reference; batch production gets a short, fast set that catches assembly faults.
| Stage | Typical acoustic checks | Purpose |
|---|---|---|
| Engineering sample (EVT) | Full response, output, distortion, leak, mic, call tests | Compare toolings and tuning options |
| Design validation (DVT) | Full set on several units, plus after sweat and drop tests | Check consistency and durability |
| Golden sample | Full set, recorded as the reference with the tuning file version | Fix the pass bands for production |
| Batch line test | Quick response, rub and buzz, mic sensitivity on every unit | Catch assembly faults |
| Pre-shipment inspection | Sampled response, leak, call check and firmware version | Confirm the lot matches the golden sample |
Sample stages and their timing are explained in the smart glasses sample process, and pre-shipment sampling in pre-shipment inspection.
Reading and using the results
Good acoustic records are short and comparable. For each unit tested, the record should show the method reference, the firmware and tuning file version, the curves or values, and pass or fail against the golden sample band. A common mistake is comparing response curves without aligning the level at a reference frequency, which makes a louder unit look as if it has more bass. Another is treating a single noisy unit as a design problem; a second and third unit from the same lot usually show whether it is a design or an assembly issue.
When batch checks drift, the usual causes are a changed mesh, a changed driver lot, a gasket issue or a firmware mismatch, and the record makes the cause easier to find. Specific test counts and pass rules are a stated working default, confirmed per order in writing.
Results should also be read together with the durability tests. A pair that measures well when new but loses low end or microphone sensitivity after artificial sweat immersion or drop tests has a sealing or mesh problem, and that is far cheaper to fix at DVT than after launch. Rechecking response and microphones after durability tests is a small addition to the plan with a large payoff.
- Agree the method before the first engineering sample.
- Run the full set on each candidate and record it against the tuning file.
- Freeze the reference on the golden sample with pass bands.
- Run line tests on every unit and sampled checks at inspection.
- Investigate drift with second and third units before changing the design.
Building tests into the program
Acoustic testing sits alongside radio, battery and optical testing in the program plan, and is coordinated with partner plants in the Shenzhen electronics base where the temples are built. Typical sample timing is about 5 to 10 working days for catalog samples, 10 to 15 for logo samples and 30 to 60 days for ODM engineering samples including tooling, confirmed in writing per program on MOQ and lead time. List your acoustic priorities on the quote form so the test method is drafted with the first sample request.