Loading 3D model… large boards can take a moment.
USB-powered ESP32-S3 ambient sound-level logger for home offices and shared rooms that keeps only privacy-preserving loudness summaries and serves a local history dashboard—never raw audio.
Quiet Trace is an open-hardware instrument for understanding when a room becomes noisy without installing a recorder or sending microphone data to a cloud service. An ESP32-S3 samples a digital MEMS microphone, calculates bounded-window loudness statistics in memory, discards each audio buffer, and stores only aggregate records such as one-minute equivalent level, peak window, and histogram buckets. A responsive dashboard served on the local network shows live level, timelines, annotations, and portable exports.
Quiet Trace is a practical survey and awareness tool, not a certified sound-level meter, occupational exposure instrument, medical device, security monitor, or legal evidence recorder. Displayed units must be labeled uncalibrated until the device has completed a documented reference-meter calibration.
Phone sound-meter apps are convenient but tie up a phone, vary by device, and can create understandable recording concerns. Professional logging meters are expensive and often export through proprietary software. Quiet Trace aims for a small, repairable, visibly non-recording device that can remain on a desk, compare rooms or time periods, and keep the data owner in control.
If Wi-Fi is unavailable, logging and physical status remain functional; USB serial supports configuration and data export. No internet service or account is required.
Raw microphone samples exist only in bounded RAM buffers required for calculation and are overwritten after each window. The device must never write or transmit samples, recordings, transcripts, or reconstructable spectral data. Stored records contain aggregates, device/session metadata, user-entered annotations, calibration metadata, and timestamps. Data stays in on-device flash and the browser’s local storage unless the user explicitly downloads an export.
The dashboard requires local-network access. Wi-Fi credentials are stored only on the device, must be erasable by a physical reset flow, and are never included in ordinary exports. Setup credentials/tokens must not be hard-coded. No microphone permission is requested from the phone or computer because the device owns the sensor. Optional browser file access is user-initiated for backup restore only.
Revision A selects an Espressif ESP32-S3-WROOM-1-N8R8 module and TDK InvenSense ICS-43434 I²S microphone, with USB-C 5 V power/protection, non-color-only LED status, physical RESET and SETUP/MARK controls, recovery header, and test access. Exact selected parts, pin/package checks, acoustic constraints, the LDO thermal bound, and point-in-time sourcing evidence are documented in hardware/parts-selection.md. Stock, prices, and formal lifecycle status still require a pre-order procurement check.
Current editable sources are:
hardware/kicad/quiet-trace.kicad_pro
hardware/kicad/quiet-trace.kicad_sch
hardware/kicad/quiet-trace.kicad_pcb
PDFs and renders may supplement these files but will never replace them. Final BOM data lives in KiCad schematic symbol properties (Manufacturer, MPN, supplier fields, and notes) and is exported to tracked bom/bom.csv. bom/preliminary-bom.csv remains planning input only.
Prototype planning target: USD 35–60, including electronics, simple enclosure, cable, and ordinary fasteners but excluding a phone/computer, tools, and a reference sound meter. This is a target rather than a live quote; the tracked BOM preserves TBD wherever current unit pricing was not actually observed, and all availability must be rechecked before ordering.
Status: issue-5 dashboard/protocol slice is merged; issue-6 physical integration evidence is in progress. The repository now contains architecture/privacy/metric contracts, selected schematic/BOM, PCB/mechanical draft, host-tested firmware domain contracts, and the merged local dashboard implementation with typed aggregate protocol fixtures. Issue #6 assembly/bring-up/calibration/troubleshooting guides are published, but real bench measurements and end-to-end physical evidence are still required before any calibrated claim. No certification, safety, legal, or bench-calibrated claim is made.
See PLAN.md, hardware/requirements.md, and the GitHub issue backlog.
Issue-#6 integration documents:
Pinned versions and setup details are in docs/toolchains.md.
# Dashboard: exact lockfile, static checks, tests, and production bundle
cd app
npm ci
npm run format:check
npm run lint
npm run typecheck
npm run test:run
npm run build
# Host-testable firmware domain contract
cd ..
cmake -S firmware/host_tests -B /tmp/quiet-trace-host-tests \
-DCMAKE_BUILD_TYPE=Release
cmake --build /tmp/quiet-trace-host-tests --parallel
ctest --test-dir /tmp/quiet-trace-host-tests --output-on-failure
# ESP-IDF v6.1 target scaffold (after exporting the pinned IDF environment)
cd firmware
idf.py set-target esp32s3
idf.py build
The dashboard’s synthetic view is opt-in at /?fixture=1 and visibly marked as fixture-only. The default route does not fabricate sensor data. Target flashing, microphone behavior, calibration, and bench results remain unverified until the exact hardware is selected and present.
The initial documentation/software scaffold is MIT licensed. Before hardware source release, the project will add an explicit open-hardware license (planned CERN-OHL-S-2.0) and document boundaries for firmware, web app, documentation, and third-party assets.
No comments yet. Be the first to ask about this board.