dspNode: Real-time remote audio rendering
summary
Proposal moves real-time audio DSP out of the browser so a lightweight client controls a cloud renderer that streams fully mixed audio back. Implementation uses Web Audio API running in node-webkit inside Docker on AWS Fargate, WebRTC with Opus stereo ~180–220 ms measured on 4G, signaling via PJS, and one rendering container spun per client. Demos show multi-channel stereo over iOS/Chrome/Firefox/Safari working in practice and cost scales per-second per-instance.
AI-written summary · about
chapters (10)
- 0:00Introduction to dspNode
- 1:06Problem with client-side audio
- 5:30WebRTC and codecs overview
- 7:19System architecture explained
- 10:12Client library design
- 12:03Latency measurement demo
- 14:32Complex audio demo
- 16:59Native audio rendering idea
- 18:43Scaling and cost analysis
- 20:05Future and Q&A
related talks
33mRealtime web applications, from your Mac to the Cloud, with SignalR and Azure
14mA Cloud Gaming Framework for Dynamic Graphical Rendering Towards Achieving
35mPushing innovation and accessibility with real-time and the cloud
18mIntegrating Telepresence with Docker Desktop: Tales from the Docker Extensions Avant-Garde
28mOpening Keynote
13mCloud XR Experience on 5G with Mobile Edge Networks
12mCloud Gaming is Real: Talk and Live Demo pt.2
07mBeyond The Browser: Wasm in Your Cloud Development Workflow