Preset: Standard Deployment {#default}
Deploy an edge-based voice collection and analysis system for your retail store.
| Device | Purpose |
|---|
| reRouter CM4 | Edge computing device, runs voice services |
| reSpeaker XVF3800 | 4-mic array for capturing store conversations |
What you'll get:
- Real-time voice transcription from store conversations
- Speaker recognition - identify who is speaking
- SenseCraft Voice cloud platform for multi-store analytics
- Privacy-first design - audio processing happens on-device
Requirements: USB-C data cable · Network cables
Step 1: Flash OpenWrt Firmware {#firmware type=manual required=false}
Write the operating system to the reRouter, then connect it to your network. Skip this step if your reRouter was purchased after November 2025 — it already has the correct firmware.
| Device | Connection | Notes |
|---|
| reRouter CM4 | Remove case to access the board | Required for entering boot mode |
| USB-C cable | Connect reRouter to computer | For eMMC flashing |
| Computer | Install rpiboot tool first | Otherwise the eMMC won't be recognized |
Prerequisites:
- Install rpiboot tool — the computer cannot recognize the eMMC without it
- Windows: Download and run rpiboot installer
- Mac/Linux: Build from source —
git clone --depth=1 https://github.com/raspberrypi/usbboot && cd usbboot && make
Flashing Steps:
-
Remove the reRouter case to access the CM4 board
-
Connect the jumper between the Boot and GND pins on the board to enter boot mode (see image below)
-
Connect reRouter to computer via USB-C, then run rpiboot — the eMMC will appear as a USB drive
-
Download firmware (you must use the links below to ensure the default IP is 192.168.49.1):
-
Flash the firmware using either tool:
-
After flashing, remove the Boot-GND jumper, reassemble the case, connect cables and power on
For detailed flashing instructions, see reRouter Flashing Guide.
First Connection:
- Connect computer to reRouter's LAN port with network cable
- Connect WAN port to your router with another network cable
- Wait 1-2 minutes for boot
- Visit
http://192.168.49.1 in browser (this is the default OpenWrt LAN IP)
- Login: username
root, password empty
Troubleshooting
| Issue | Solution |
|---|
| Cannot access 192.168.49.1 | 1) Make sure cable is in the LAN port; 2) Make sure you flashed the firmware from the links above (other firmware may use a different IP) |
| Page loads slowly | Wait 2 minutes for the system to fully boot |
| rpiboot doesn't detect the device | Make sure the Boot-GND jumper is connected; try a different USB-C cable |
| Flashing failed | Format the storage device and try again |
| Login failed | Password is empty, just click login |
Step 2: Deploy Voice Services {#voice_services type=docker_deploy required=true config=devices/rerouter.yaml}
Start the voice recognition and analysis services on the device.
Target: Local Deployment {#voice_services_local type=local config=devices/voice_local.yaml}
Deploy voice services on your local computer.
Wiring
| Device | Connection | Notes |
|---|
| reSpeaker XVF3800 | USB to computer | Make sure it's a data cable, not just a charging cable |
| Computer | Docker Desktop installed | Download for Windows/Mac |
- Make sure Docker Desktop is installed and running
- Confirm reSpeaker XVF3800 is connected via USB
- Verify at least 2GB free disk space and port 8090 is available
- Check reSpeaker is recognized: Windows Device Manager > Sound controllers; Mac System Preferences > Sound > Input; Linux run
arecord -l
Troubleshooting
| Issue | Solution |
|---|
| Docker not running | Start the Docker Desktop application |
| Port 8090 is occupied | Close the program using that port, or modify the configuration to use a different port |
| Microphone device not found | Unplug and replug USB, verify it appears in Device Manager |
| Container startup failed | Check Docker logs: docker logs sensecraft-voice-client |
Target: Remote Deployment {#voice_services_remote type=remote config=devices/rerouter.yaml default=true}
Deploy voice services to a remote device (reRouter, Raspberry Pi, etc.).
Wiring
| Device | Connection | Notes |
|---|
| reSpeaker XVF3800 | USB to reRouter | Audio settings are configured automatically during deployment |
| reRouter CM4 | WAN port to router | Internet required for downloading container images |
| reRouter CM4 | LAN port to computer | For SSH access and deployment |
| Computer | Same network as reRouter | For running remote deployment |
- Confirm reRouter WAN port is connected to router and has internet
- Connect computer to reRouter LAN port
- Default SSH: IP
192.168.49.1, user root, no password
- Plug reSpeaker XVF3800 into reRouter USB port
Troubleshooting
| Issue | Solution |
|---|
| SSH connection refused | Make sure the cable is plugged into the LAN port and the IP is correct |
| Authentication failed | OpenWrt default password is empty, just press Enter |
| Image download timeout | Check the WAN port network connection, make sure you can access the internet |
| Container startup failed | SSH in and run docker logs sensecraft-voice-client to view error messages |
| Microphone not found | Run arecord -l to verify reSpeaker is recognized |
| "Health check failed" warning in logs | Normal during startup - the voice client starts before the ASR server is ready. Wait 30 seconds and check again |
Deployment Complete
Voice AI system is ready!
Service Access
After deployment, you can access the following services:
| Service | URL | Purpose |
|---|
| Edge Client | http://<device-ip>:8090 | Real-time transcription, speaker management, device settings |
| OpenWrt Admin | http://<device-ip> | Network configuration, system management |
| SenseCraft Voice | https://test-voice-web.seeed.cn/ | Cloud platform - multi-store analytics, AI analysis, data export |
Initial Setup
- Reboot the device — run
reboot command via SSH, then wait 2 minutes
- Open Edge Client — visit
http://<device-ip>:8090 in your browser
- Test voice recognition — speak near the reSpeaker and watch real-time transcription appear
Edge Client (http://<device-ip>:8090)
The edge client provides local voice processing and device management:
| Feature | Description |
|---|
| Real-time ASR | Live speech-to-text transcription - verify audio input and recognition accuracy |
| Speaker Recognition | Register voiceprints to automatically identify who is speaking |
| Device Configuration | Change network settings (WiFi) and upstream server address for cloud sync |
Connect your edge device to the SenseCraft Voice cloud platform for advanced analytics:
| Module | Description |
|---|
| Dashboard | Overview with store filtering, daily collection trends, and keyword hotspot analysis |
| Record Management | Search, filter, and export voice records. Switch between conversation view and timeline view with audio playback |
| AI Analysis | Submit voice records to AI for custom processing based on your prompts |
| Store Management | Organize devices by store, location, and device name |
| Backend Settings | Configure keywords & synonyms for event detection, AI prompts, and user permissions |
To connect to the cloud platform:
- Open Edge Client > Device Status page
- The upstream server address is pre-configured
- Your device will automatically register and appear in the cloud platform
Quick Verification
- Speak near the reSpeaker microphone
- Check edge client for real-time transcription
- Verify text appears on the web dashboard
- Check speaker recognition identifies different speakers
Next Steps
