Preset: Standard Deployment {#default}

Deploy an edge-based voice collection and analysis system for your retail store.

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.

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:

1. Remove the reRouter case to access the CM4 board

2. Connect the jumper between the Boot and GND pins on the board to enter boot mode (see image below)

   

3. Connect reRouter to computer via USB-C, then run rpiboot — the eMMC will appear as a USB drive

4. Download firmware (you must use the links below to ensure the default IP is 192.168.49.1):

   • Global | China

5. Flash the firmware using either tool:

   • Raspberry Pi Imager — select "Use custom" and choose the firmware
   • balenaEtcher — select firmware file and target drive, click "Flash"

6. 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:

1. Connect computer to reRouter's LAN port with network cable
2. Connect WAN port to your router with another network cable
3. Wait 1-2 minutes for boot
4. Visit http://192.168.49.1 in browser (this is the default OpenWrt LAN IP)
5. Login: username root, password empty

Troubleshooting

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

1. Make sure Docker Desktop is installed and running
2. Confirm reSpeaker XVF3800 is connected via USB
3. Verify at least 2GB free disk space and port 8090 is available
4. Check reSpeaker is recognized: Windows Device Manager > Sound controllers; Mac System Preferences > Sound > Input; Linux run arecord -l

Troubleshooting

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

1. Confirm reRouter WAN port is connected to router and has internet
2. Connect computer to reRouter LAN port
3. Default SSH: IP 192.168.49.1, user root, no password
4. Plug reSpeaker XVF3800 into reRouter USB port

Troubleshooting

Step 3: User Onboarding Guide {#user_guide type=manual required=false}

Verify your system is working and connect it to the cloud platform.

Step 1: Open Edge Client and Start Recording

1. Open http://<device-ip>:8090 in your browser (default: http://192.168.49.1:8090)

   

2. Click the Record button — speak near the reSpeaker and watch transcribed text appear in real time

3. Navigate to the Device Status page in the sidebar

   

4. The upstream server address is pre-configured — your device will auto-register to the cloud platform once it can reach the server

Step 2: Find Your Device on the Cloud Platform

1. Open https://test-voice-web.seeed.cn/

2. Go to Store Management — locate your device by its IP address or MAC address

   

3. Click your device to take control — all transcribed voice records are now available under Record Management

Troubleshooting

Deployment Complete

Voice AI system is ready!

Verify Deployment

After deployment, confirm all services are running:

# Check container status — all three should show "Up"
docker ps

# Check voice client initialization logs
docker logs sensecraft-voice-client

Then reboot the device to ensure all settings and audio permissions take effect:

reboot

Wait 2 minutes after reboot before proceeding.

Service Access

Edge Client (http://<device-ip>:8090)

The edge client runs locally on the reRouter and provides three modules:

Voice ASR

Shows the real-time operational status of the on-device ASR service. Speak near the reSpeaker and watch transcription appear instantly — use this to verify audio input is working and recognition accuracy meets your needs.

Voiceprint Recognition

Register speaker voiceprints so the system can automatically identify who is speaking in recordings. The system builds unique voiceprints from audio samples — once registered, speakers are labeled automatically in all future transcriptions.

Device Status & Configuration

View reRouter operating status and adjust core parameters:

• Network settings — configure Wi-Fi connection
• Upstream server address — the address for syncing data to the cloud platform (pre-configured, change only if using a private deployment)

Cloud Platform (https://test-voice-web.seeed.cn/)

Connect your edge device to the SenseCraft Voice cloud platform for multi-store analytics. Your device will auto-register once the upstream server address is set in the Edge Client.

Dashboard

Operational overview with store filtering — switch between stores and see charts update instantly. Monitors daily collection trends by hour and surfaces keyword hotspot analysis showing which keywords triggered most frequently and from which devices.

Record Management

Search and filter voice records by device name, store name, location, or MAC address. Two viewing modes:

• Conversation Mode — review transcribed dialogues
• Timeline Mode — play back original audio alongside transcription

Export records in three formats: Markdown, Plain Text (.txt), or original audio files.

AI Analysis

Submit filtered voice records to AI for custom processing. The analysis history is stored chronologically so you can review previous sessions. Only one AI prompt is active at a time — configure which prompt to use in Backend Settings.

Store Management

Organize your deployment in a three-level hierarchy: Store → Location → Device. Group devices logically to simplify large-scale management — filter records by any level of this hierarchy.

Backend Settings

Configure the system's detection and processing behavior:

Keyword Settings

Define custom keywords and synonyms for event detection. Assign a color to each keyword for visual highlighting on the dashboard. Supports batch add, edit, and delete.

AI Prompt Settings

Create custom prompts that control how the AI processes and summarizes voice records. Only one prompt can be active at a time — enable/disable as needed for different analysis tasks.

User Management

Control who can access the cloud platform and manage their permissions.

Next Steps

• View Wiki Documentation
• SenseCraft Voice Platform
• Purchase Hardware