Key Technologies and Applications of Tencent Real-Time Audio‑Video (TRTC)

With the development of mobile Internet, audio‑video services have shifted from one‑way viewing to multi‑party interaction, demanding lower latency and richer interactivity. This article summarizes the sharing of Jiang Lei, a senior engineer of Tencent Cloud Real‑Time Audio‑Video (TRTC), at a Cloud+ Community offline salon, and analyzes the core technologies and applications of TRTC.

1. Evolution of Internet Communication Services – From traditional mail, SMS, and phone/fax to IM‑based communication (QQ, WeChat) and now to Communication 3.0 (industry Internet) where human‑to‑human, human‑to‑device, and service‑to‑human connections are all required. Real‑time audio‑video (RTC) is the key enabler.

Typical RTC scenarios include online meetings, video customer service, online education, live board‑games, K‑song, voice chat rooms, FM radio, and live streaming. Tencent’s TRTC also supports many other custom scenarios.

2. Tencent Real‑Time Audio‑Video (TRTC)

TRTC is a PaaS service sold on Tencent Cloud, providing cross‑platform SDKs for mobile, desktop, web, and mini‑programs. During the pandemic, TRTC supported massive online education and remote work, reaching a daily traffic of 30 billion minutes, making it one of the world’s largest RTC services.

Key use cases:

Multi‑party real‑time interaction (e.g., Tencent Meeting) with end‑to‑end latency under 300 ms and support for 300 participants per room, up to 30 simultaneous speakers.

Low‑latency live streaming (latency < 1 s) for platforms such as Tencent Classroom, Enterprise WeChat, and WeChat Video Channels.

3. How Real‑time is Achieved

Achieving sub‑second latency (300‑1000 ms) requires overcoming network loss, jitter, and variable bandwidth. TCP’s retransmission and buffering introduce high latency, while UDP with NACK, FEC, and ARQ reduces delay.

TRTC uses UDP with NACK to request only lost packets, and applies Forward Error Correction (FEC) to send redundant packets, reducing the need for retransmission. QoS predicts bandwidth and adjusts encoding rates dynamically.

Video codec modifications relax strict I‑frame dependencies, allowing selective reference frames to mitigate the impact of packet loss. Audio uses Packet Loss Concealment (PLC) to reconstruct missing frames based on temporal correlation.

Echo cancellation (AEC) and dual‑talk handling are performed by Tencent Real‑time Audio Engine (TRAE), eliminating self‑echo and improving clarity in full‑duplex scenarios.

4. Deep Optimizations

Audio‑video synchronization is achieved by jitter assessment and dynamic speed adjustment of audio streams to align with video frames.

A cloud‑based decision system (QoE) collects real‑time quality metrics from clients, evaluates user experience, and dynamically tunes parameters per user, ensuring optimal performance even under diverse network conditions.

Scalable live streaming has been extended to support up to 100 000 concurrent users in a single room, with optional CDN off‑loading for even larger audiences.

5. Additional Capabilities

TRTC SDKs are integrated into WeChat Mini‑Programs via <live-pusher> and <live-player> tags, offering native‑like latency and audio quality.

Advanced features include AI‑driven beauty filters, background replacement, voice effects (voice change, reverb), and a comprehensive monitoring dashboard that tracks end‑to‑end network and device health.

6. Customer Cases

TRTC has been deployed for over a thousand customers across finance, IoT, automotive, real‑estate, and many other industries, providing real‑time audio‑video capabilities for virtual bank counters, smart‑car showrooms, VR house tours, and large‑scale live events.