From smart voice hardware and the voice dialogue platform to the IoT smart gateway and multi-platform apps, one product line covers the full chain of “hear, understand, respond and connect”. Each of the four product lines can be delivered on its own or combined into a complete device solution.
ESP32-based smart voice hardware — plug and play
A desktop voice terminal that is ready the moment it wakes, with continuous conversation and barge-in. Wake-word detection and basic commands run on the device, while complex dialogue is handled by cloud LLMs — balancing response speed with conversational ability.
Wake-word detection triggers automatic sentence segmentation, with multi-turn conversation and barge-in — no need to re-trigger the wake word every time.
Connect to mainstream LLMs and voice services such as GLM, Qwen and iFlytek, and switch per scenario — no lock-in to a single vendor.
BLUFI Bluetooth provisioning: bring your phone close to finish network setup, with no tedious Wi-Fi password entry.
Wake word and basic commands are processed locally for responsiveness, while complex semantics and knowledge Q&A go to the cloud. Basic interaction still works when the network is down.
| Main control | ESP32 series — local wake-up plus cloud dialogue, dual-path processing |
|---|---|
| Audio | Microphone capture and speaker output, with echo cancellation so audio keeps flowing while playing |
| Network | Wi-Fi connectivity plus Bluetooth BLE, supporting Bluetooth provisioning and Bluetooth audio |
| Power | USB-C powered, desktop placement |
| Form factor | Desktop or wall-mounted; enclosure, cutouts and connector customization available |
| Upgrades | OTA remote firmware updates, with continuous feature iteration |
Full-chain AI voice service — the brain behind the hardware
Speech recognition, LLM dialogue, speech synthesis, intent understanding and memory management are strung into one callable service, so devices, apps, mini programs and enterprise systems share the same interface.
ASR → LLM dialogue → TTS are connected end to end, with intent understanding and memory management built in — no need to assemble multiple vendors yourself.
Switch between multiple LLMs and voice services behind one interface, with automatic failover to avoid single points of failure.
Streaming across the whole chain: recognize while speaking, generate and play back as you go — noticeably shortening the wait from finishing your sentence to hearing a reply.
Multi-turn context and long-term memory management, with business systems plugged in as skills so the assistant can actually get things done.
| Interface | HTTP / WebSocket, compatible with OpenAI-style chat APIs for easy integration |
|---|---|
| Capabilities | Speech recognition, LLM dialogue, speech synthesis, intent understanding, memory management, tool calling |
| Models | Mainstream LLMs and voice services such as GLM, Qwen and iFlytek, with room to extend |
| Deployment | Public cloud API access, or private deployment for enterprise customers |
| Consumers | Device firmware, Android / iOS apps, WeChat mini programs, enterprise business systems |
| Management | Web admin console for centralized device and firmware management |
Turning voice commands into real device action
An MQTT gateway and provisioning solution for smart devices: device onboarding, online management and command delivery — turning "say one sentence" into "the device really moved".
A unified channel for message intake and delivery, with persistent connections and automatic reconnection after disconnects.
A Bluetooth provisioning approach with a stable, reliable setup process, suited to screenless smart devices.
Visualized management of online status, command delivery and status reporting — anomalies are traceable.
Voice commands drive device control directly, with scenario automation and industry-specific custom commands.
| Connectivity | Wi-Fi / Bluetooth, MQTT protocol |
|---|---|
| Provisioning | Bluetooth provisioning (BLUFI), no screen interaction required |
| Automation | Voice command → device control, with scenario orchestration and custom commands |
| Extension | Relays, sensors and other modules can connect lights, curtains, sockets and more |
| Upgrades | The gateway and sub-devices all support OTA remote upgrades |
Beyond the device — users and operations stay online
Mobile apps and mini programs let users bind, talk and upgrade; the Web admin console lets operators manage devices and firmware centrally.
Device binding, a voice interaction entry point, usage history and personal settings, with a consistent experience across both platforms.
An install-free entry point — scan and use — suited to demos, events and lightweight scenarios.
Detect and deliver firmware updates inside the app, with staged rollouts to reduce upgrade risk.
Centralized device and firmware operations, with status views, version management and troubleshooting.
| Clients | Android / iOS apps, WeChat mini program |
|---|---|
| User capabilities | Device binding, voice interaction, firmware upgrades, usage management |
| Admin console | Web console for centralized device and firmware operations |
| Upgrades | Both apps and device firmware support online upgrades |
Need a sample unit, a module or a custom solution? Reach us through Contact and we will get back to you shortly.