How a 33‑Hour Continuous 4K + Dolby Atmos Live Stream Was Engineered

Project Background

In Bilibili Music’s 2024 concert series, the first event was Taylor Swift’s Singapore show. The team delivered a 33‑hour uninterrupted live stream on the Bilibili platform, attracting over 10 million popularity points.

End‑to‑End 4K + Dolby Atmos Solution

The production achieved full‑chain 4K 59.94 P video and Dolby Atmos audio. A dedicated SD‑WAN link connected the Singapore remote studio to the Beijing Dolby Atmos studio, allowing bidirectional transmission of audio‑video signals.

Dolby Atmos Overview

Dolby Atmos, originally a cinema technology, provides immersive sound without channel limits. Mix engineers can place and move individual sounds in a 3‑D space, controlling volume, size, and spatial characteristics for fine‑grained detail.

Rapid Deployment

From specification to execution, the team completed the project in just nine days, marking its first high‑spec music live‑stream with Dolby Atmos despite prior experience mainly in events and product launches.

Core Devices

Video: 1 × Blackmagic Design ATEM 4 M/E Constellation 4K, 2 × ATEM Constellation 8K, 1 × Blackmagic Videohub 80x80 12G, 3 × Sony FX6, 2 × Canon XF605.

Audio: 1 × GENELEC 8351B+7370A 9.1.6 Dolby Atoms Studio, 1 × Solid State Logic System T S300.

Transmission: 2 × Ateme Titan Live, 2 × Blackmagic Web Presenter 4K, 2 × AVIWEST PRO460, 1 × AVIWEST PRO3‑5G.

System Overview

The Beijing studio, located in a Dolby Atmos‑equipped soundstage, received video and audio from Singapore via the SD‑WAN link. All links used industry‑standard 4K 59.94 P signals, delivering richer channel detail than stereo.

Video System Details

The main switcher was a Blackmagic Design ATEM 4 M/E Constellation 4K handling nine 4K 59.94 P signals (various camera angles and playback servers). An 80‑input/80‑output Blackmagic Videohub matrix ensured reliable switching.

Transmission Path

Singapore’s remote studio used AVIWEST 5G backpacks, encoding at 4K 20 Mbps HEVC and 1080P 10 Mbps HEVC, achieving end‑to‑end 2000 ms latency over a public 5G network.

MV Playback and Synchronisation

Audio and video were played back separately and synchronised via timecode. QLab controlled audio playback and triggered the PVP video server and vMix subtitle overlays.

Dual Output Streams

Two versions were generated for audience devices: a 4K + Dolby Atmos stream and a 4K + stereo stream, allowing Bilibili to deliver the appropriate signal based on each device’s decoding capability.

Recording and Encoding

Three Blackmagic HyperDeck Studio 4K Pro units recorded the video (two for Dolby, one for stereo). Encoding used two Ateme Titan Live servers for the Dolby version and two Blackmagic Web Presenters for the stereo version.

Audio System Architecture

The audio chain centred on an SSL TE2 DSP processor with 800 channels, a 32‑fader System T S300 console, and a native 7.1.4 bus. All audio ran over Dante AoIP, inter‑connected via two Cisco SG300‑28SFP switches and an Avid Mtrx II Dante‑to‑analog converter.

Dolby Atmos Production

The system used a Dolby Atmos 5.1.4 base format, with flexible 3D panning and immersive compressors/EQ. Down‑mix functionality produced a stereo version when needed.

Field Capture in Singapore

Sennheiser AMBEO VR Mic captured ambient sound, feeding four XLR channels into the SDI signal. The 5G backpack transmitted these alongside video. SSL’s 360 Transcode plugin decoded the signals to Dolby Atmos, and guest mics were mixed locally before being sent back.

Audio‑Video Integration

Blackmagic switcher’s MADI embed/de‑embed converted between Dante IP audio and baseband MADI.

Focusrite Rednet D64R and RME Digiface Dante handled conversion, supporting 12‑channel front‑of‑house and CG audio, plus 10‑channel Dolby and 2‑channel stereo embed.

Lawo V__pro8 served as a backup embed/de‑embed device, enhancing reliability.

Audio Quality Control

Dolby DP580 decoders monitored the Atmos stream, simulating stereo, 5.1, and immersive playback to ensure consistent high‑quality experience across all audience devices.

Intercom System

RTS Zeus III matrix with multiple RTS Keypanels connected to a Mac Mini running Unity Intercom enabled multi‑party communication and an IFB channel for guest earpieces.

Network Infrastructure

A 100 Mbps SD‑WAN link linked Singapore and China, while the Beijing studio used a 2 Gbps dedicated internet line with dual encoders on separate ISP routes for redundancy.

Staff and Acknowledgements

The production involved a large technical crew covering video, audio, intercom, networking, and post‑production roles.