Tencent Car‑Mounted Mini Program Architecture and Voice Interaction

Tencent's car‑mounted mini program is a smart travel solution that brings convenient services to users in vehicles and supports voice control. It is compatible with WeChat mini programs, allowing developers to adapt existing mini programs to run on car head‑units with minimal effort.

The presentation explains how developers can create content and services for intelligent travel scenarios using mini programs.

The overall architecture is divided into three layers:

1. Bottom layer – Tencent Browser Service (TBS) : provides a JavaScript runtime and rendering engine for executing mini programs.

2. Middle layer – Mini program framework with TAIS voice interaction : extends the standard WeChat mini program framework by adding voice capabilities, enabling hands‑free operation in the car.

3. Top layer – Mini program applications : a variety of services (travel, audio‑visual, lifestyle) delivered through the framework.

From the client side, additional features include voice recording upload, noise reduction, vehicle status integration via push channels, and security policies specific to the car environment.

On the cloud side, a management backend handles voice recognition, semantic processing, recommendation logic, and safety checks before launching a mini program.

The design emphasizes three core benefits:

Lightweight, rich service ecosystem that leverages the massive existing WeChat mini program developer base.

Safe and convenient voice interaction, reducing the need for manual input while driving.

Intelligent, scenario‑based recommendation (e.g., location‑based traffic alerts, vehicle‑maintenance prompts, nearby points of interest). UI adaptation is minimal: the framework can automatically adjust navigation bars and layout for the car's horizontal screen, and responsive CSS techniques enable the same mini program to run on phones, tablets, and car displays. Two presentation modes are supported: full‑screen and floating overlay (useful when a mini program needs to appear over navigation). Developers can add voice command attributes to UI controls (e.g., a payment button with trigger words like "pay" or "checkout") and define semantic skills that map spoken intents to functions. Two interaction flows are described: Voice‑triggered launch : the driver issues a voice command, the car’s backend parses intent, selects an appropriate mini program (e.g., JD for buying water), and launches it directly to the relevant page. Smart recommendation launch : the car’s telematics system pushes contextual events (traffic accidents, maintenance reminders) to the backend, which then pushes a mini program notification to the driver for optional activation. Safety checks are performed before launching any mini program to ensure it does not distract the driver. The solution supports multiple car OS platforms (Android, Linux, QNX), allowing a single mini program to serve various vehicle manufacturers. Overall, the car‑mounted mini program aims to be install‑free, instantly available, and seamlessly integrated into the driving experience, providing a rich, safe, and intelligent service ecosystem.