-
Notifications
You must be signed in to change notification settings - Fork 0
Dashboard
Access the dashboard at http://<server-ip>:8091 (or https://soundspy.silver.local if Traefik is configured via the override file).
Each active node gets a card showing:
- Overall level meter (dBA by default, click the unit label to toggle to dBFS)
- 6-band EQ spectrum — switchable between bar and wave visualization modes
- Sleep/wake state indicator
- Firmware version and IP address
Nodes are auto-discovered: any ESP32 publishing to soundspy/<chip_id>/data appears automatically without any manual configuration.
New nodes are auto-named node-1, node-2, etc. Click the name in a node card to rename it inline. Names persist across restarts in data/node_names.json.
The dashboard streams live PCM audio from each node via WebSocket. Click the audio icon on a node card to listen. Latency is under 100 ms. Audio has gain applied (same as the published level data).
Per-node controls available from the dashboard:
- Gain slider — adjusts capture gain (0–10×) sent to the node via MQTT
- Sleep / Wake — stops I2S on the node to reduce power; MQTT remains connected so the node stays reachable
- OTA update — upload a firmware binary and push it to a specific node wirelessly
- Reboot — sends a reboot command via MQTT
Each node maintains a 1-hour in-memory history (3600 samples). The history chart supports selectable time ranges. Level values are averaged in the linear power domain before display to preserve perceptual accuracy.
- Bar mode — standard spectrum bar chart
- Wave mode — continuous waveform overlay
Both chart types use colored background zones:
- Green (−80 to −20 dBFS): quiet / ambient
- Yellow (−20 to −10 dBFS): noticeable
- Red (−10 to 0 dBFS): loud / clipping
A logs panel shows remote log messages from all nodes, split into per-node sidebars and an app/system tab. Firmware publishes log events over MQTT to soundspy/<chip_id>/log and boot reports to soundspy/<chip_id>/boot.
The threshold monitor service watches the soundspy/+/data topic and sends an ntfy push notification when bass exceeds the configured threshold. Default threshold is −20 dBA with a 5-minute cooldown. Configure via FREQ_THRESHOLD_DBFS and COOLDOWN_SECONDS in .env.
The dashboard displays levels in dBA by default. dBA is computed by applying A-weighting offsets to the per-band dBFS values. The A-weighting offset for each band is cached after the first calculation. Click the unit label on any level display to toggle between dBA and dBFS.
The alert threshold is stored internally in dBFS and defaults to −20 dBFS (approximately −20 dBA for typical broadband noise). It is set via FREQ_THRESHOLD_DBFS in .env and is applied by threshold_monitor.py.
Each node card shows a threshold line on the level meter. The position reflects the dBFS value; the displayed label switches with the dBFS/dBA toggle.
Increase gain if the displayed levels seem too low for your environment. Decrease it if the meter is clipping (reaching 0 dBFS) during normal use. Gain changes take effect immediately on the node; no rebuild or reflash is needed.