WebRTC Live Castos https://webrtc.ventures/webrtc-live/ WebRTC Live is a monthly podcast from WebRTC.ventures, hosted by Arin Sime, recorded during our live webinar broadcast. Engineers and founders discuss how real-time communications and voice AI applications are built and run in production, covering open source media servers and CPaaS, VoIP and SIP integration, WebRTC monitoring and debugging, protocols such as MoQ and WebTransport, and AI agents, avatars, and video analysis. Episodes also look at specific use cases, including telephony and contact centers, telehealth, telecoms, broadcasting, live collaboration, and more. Past guests include engineers and leaders from OpenAI, Cloudflare, LiveKit, and Meetecho. Wed, 04 Nov 2026 19:02:23 +0000 en © 2026 https://episodes.castos.com/6282a1833aa061-12170476/images/podcast/covers/c1a-4o398-jppdzmj5i43d-hbpu7n.png WebRTC Live https://webrtc.ventures/webrtc-live/ WebRTC Live is a monthly podcast from WebRTC.ventures, hosted by Arin Sime, recorded during our live webinar broadcast. Engineers and founders discuss how real-time communications and voice AI applications are built and run in production, covering open source media servers and CPaaS, VoIP and SIP integration, WebRTC monitoring and debugging, protocols such as MoQ and WebTransport, and AI agents, avatars, and video analysis. Episodes also look at specific use cases, including telephony and contact centers, telehealth, telecoms, broadcasting, live collaboration, and more. Past guests include engineers and leaders from OpenAI, Cloudflare, LiveKit, and Meetecho. Arin Sime episodic WebRTC Live is a monthly podcast from WebRTC.ventures, hosted by Arin Sime, recorded during our live webinar broadcast. Engineers and founders discuss how real-time communications and voice AI applications are built and run in production, covering open source media servers and CPaaS, VoIP and SIP integration, WebRTC monitoring and debugging, protocols such as MoQ and WebTransport, and AI agents, avatars, and video analysis. Episodes also look at specific use cases, including telephony and contact centers, telehealth, telecoms, broadcasting, live collaboration, and more. Past guests include engineers and leaders from OpenAI, Cloudflare, LiveKit, and Meetecho. Arin Sime arin@agilityfeat.com false https://feeds.castos.com/jnqk0 yes <![CDATA[WebRTC, MoQ, or Both? A QUIC Study]]> Wed, 04 Nov 2026 19:02:23 +0000 Arin Sime https://permalink.castos.com/podcast/72791/episode/2691004 In the ongoing debate around WebRTC vs. MoQ, the most practical answer may be, “why not both?” The IETF is now examining whether QUIC offers real-time media applications any advantage over WebRTC, with WebTransport and MoQ positioned as possible foundations for the next generation of real-time protocols.

In this episode of WebRTC Live, we welcome back Lorenzo Miniero, Chairman at Meetecho and author of the Janus WebRTC Server. Lorenzo has been experimenting with MoQ and QUIC for several years, looking at MoQ from the perspective of a WebRTC developer. He’ll talk about building the imquic library and adding QUIC-based protocols to Janus, including RTP over QUIC (RoQ) and MoQ. We’ll discuss the problems he ran into integrating these protocols with an existing WebRTC server, the gaps between WebRTC and QUIC-based frameworks, and the right use cases for WebRTC, MoQ, or both.

The episode includes our monthly WebRTC Industry Update with Tsahi Levent-Levi. This month’s topic: Deepgram Acquires Daily: Was It a Bet on Pipecat?

Chapters

  • (00:00:00) - Introduction and Episode Overview
  • (00:00:42) - Announcements: Blog, VoiceAI Events, Guest Intro
  • (00:03:42) - Lorenzo Mineiro Joins: Intro to WebRTC and MOQ
  • (00:04:18) - Why MOQ? Origins and Early Experiments
  • (00:05:56) - MOQ Use Cases: Broadcasting, Conferencing, Synchronization
  • (00:09:12) - Configurable Latency and Live/Past Objects
  • (00:11:34) - Real-World MOQ Use Cases: Remote Driving, Gaming, Sensors
  • (00:14:24) - Viewer Question: Peer-to-Peer and SFU with MOQ
  • (00:16:38) - MOQ Relay vs WebRTC SFU
  • (00:17:39) - Congestion Control: QUIC Datagrams vs Streams
  • (00:24:12) - Building MOQ Support into Janus: Learning QUIC from Scratch
  • (00:28:31) - Bridging WebRTC and MOQ: Coexistence Philosophy
  • (00:31:50) - WebRTC vs MOQ Cheat Sheet: Transport and Signaling
  • (00:35:24) - Catalogs, Tracks, and LOQ Packaging Format
  • (00:38:21) - RTP over QUIC (RoQ) Explained
  • (00:42:24) - RoQ Use Cases: Trunking and Cascading
  • (00:46:39) - Fun Segment: Heavy Metal Bands as Media Protocols
  • (00:52:39) - Real-World Deployment: IETF Meetings with WebRTC and MOQ
  • (00:55:11) - Closing Thoughts and Where to Learn More
  • (00:56:46) - Transition to Industry Analysis Segment
  • (00:57:46) - Deepgram Acquires Daily: Initial Reactions
  • (00:58:47) - Pipecat Foundation and Open Source Concerns
  • (01:01:06) - Comparing Daily vs LiveKit: Funding and Headcount
  • (01:07:15) - Foundation Governance: Matrix/Element Comparison
  • (01:09:59) - GitHub Stars Analysis: Pipecat vs LiveKit vs Vision Agents
  • (01:12:31) - Market Implications and Deepgram's Strategic Position
  • (01:14:38) - Closing Remarks and Sign-off
]]> full 119 false In the ongoing debate around WebRTC vs. MoQ, the most practical answer may be, “why not both?” The IETF is now examining whether QUIC offers real-time media applications any advantage over WebRTC, with WebTransport and MoQ positioned as possible foundations for the next generation of real-time protocols.

In this episode of WebRTC Live, we welcome back Lorenzo Miniero, Chairman at Meetecho and author of the Janus WebRTC Server. Lorenzo has been experimenting with MoQ and QUIC for several years, looking at MoQ from the perspective of a WebRTC developer. He’ll talk about building the imquic library and adding QUIC-based protocols to Janus, including RTP over QUIC (RoQ) and MoQ. We’ll discuss the problems he ran into integrating these protocols with an existing WebRTC server, the gaps between WebRTC and QUIC-based frameworks, and the right use cases for WebRTC, MoQ, or both.

The episode includes our monthly WebRTC Industry Update with Tsahi Levent-Levi. This month’s topic: Deepgram Acquires Daily: Was It a Bet on Pipecat?

]]> 01:16:16 <![CDATA[Building the Data Layer for Voice AI with vCon]]> Wed, 14 Oct 2026 20:22:00 +0000 Arin Sime https://permalink.castos.com/podcast/72791/episode/2660935 https://webrtc.ventures/2026/10/watch-webrtc-live-118-building-voiceai-data-layer-vcon/ Voice AI applications need a way to store conversation data that supports observability, auditing, and long-term retention. The audio recording or a text transcript alone leaves out critical context. Additional details matter too, such as who was in the conversation, what devices were used, what was said, and how decisions were made.

vCons act as the “memory” of a conversation. They are structured, portable, machine-readable objects that capture conversation data across voice, video, messaging, and email, regardless of platform. This matters especially for Voice AI in regulated industries, like healthtech or fintech, where each conversation needs to be fully auditable.

Our guest is Jeff Pulver, VoIP industry pioneer and founder of the vCon Foundation. We cover the need for vCons, the current status of the IETF draft, where adoption stands, use cases and examples, how AI is changing internet communications, and more. 

The episode include our monthly WebRTC Industry Update with Tsahi Levent-Levi. This month's topic: Chrome's Neural Echo Cancellation Leaves Edge, Electron, and Native Apps Behind.

Chapters

  • (00:00:07) - Introduction and Episode Preview
  • (00:00:41) - Background on Jeff Pulver and VCONs Topic
  • (00:02:44) - Interview Begins: What is a VCON?
  • (00:03:18) - VCONs as Robot Food for AI and Memory/Trust
  • (00:07:30) - VCON Foundation, Standards, and Use Cases
  • (00:08:56) - Regulatory Issues and Consent in AI Calls
  • (00:18:21) - The Pulver Order Origin Story
  • (00:22:11) - 20th Anniversary of the Pulver Order
  • (00:23:52) - Agent-to-Agent Communication and Future of VCONs
  • (00:25:32) - Upcoming VCON Conference and Events
  • (00:35:16) - IETF Standardization Status and Timeline
  • (00:36:53) - Emotional and Personal Value of Conversations
  • (00:40:38) - Memory Care and Closing Thoughts
  • (00:42:23) - Episode Wrap-up and Next Episode Preview
  • (00:43:08) - Google's Neural Network Echo Cancellation Discovery
  • (00:45:54) - Browser Fragmentation and Quality Disparities
  • (00:51:18) - Implications for Developers and CPaaS Vendors
  • (00:52:29) - AEC3 Maintenance and Voice AI Implications
  • (00:55:53) - Closing Remarks and Sponsor Message
]]> full 118 false Voice AI applications need a way to store conversation data that supports observability, auditing, and long-term retention. The audio recording or a text transcript alone leaves out critical context. Additional details matter too, such as who was in the conversation, what devices were used, what was said, and how decisions were made.

vCons act as the “memory” of a conversation. They are structured, portable, machine-readable objects that capture conversation data across voice, video, messaging, and email, regardless of platform. This matters especially for Voice AI in regulated industries, like healthtech or fintech, where each conversation needs to be fully auditable.

Our guest is Jeff Pulver, VoIP industry pioneer and founder of the vCon Foundation. We cover the need for vCons, the current status of the IETF draft, where adoption stands, use cases and examples, how AI is changing internet communications, and more. 

The episode include our monthly WebRTC Industry Update with Tsahi Levent-Levi. This month's topic: Chrome's Neural Echo Cancellation Leaves Edge, Electron, and Native Apps Behind.

]]> 00:56:50 <![CDATA[Building Robust Voice AI Pipelines for Every Use Case]]> Wed, 12 Aug 2026 22:44:00 +0000 Arin Sime https://permalink.castos.com/podcast/72791/episode/2668832 https://webrtc.ventures/2026/08/watch-webrtc-live-116-building-robust-voice-ai-pipelines-with-varun-singh-2/ Human speech is often unstructured and messy. To keep up with real conversation, Voice AI bots must parse partial sentences, filler words, and interruptions, all without the visual cues a human listener relies on. On top of that linguistic challenge, real-world latency budgets and network conditions test every Voice AI pipeline, and users don’t forgive one that stutters.

In this episode of WebRTC Live, host Arin Sime talks with Dr. Varun Singh, Chief Product Technology Officer at Daily.co / Pipecat.ai, about the engineering decisions that separate voice agents that succeed from ones that get abandoned. Topics include where Voice AI is working today and where it isn’t, from well-defined tasks like collections to the open-ended conversations everyone is still chasing, and how Pipecat approaches the problem: composing voice agents from STT, LLM, TTS, and everything in between, fitting in speech-to-speech models, and getting from prototype to production. Also discussed: why WebRTC remains the right transport for real-world networks, the production realities of observability, cost, and failure modes, how to weigh build-vs-buy, and where voice-native UIs are headed.

The episode includes our monthly WebRTC Industry Update with Tsahi Levent-Levi. This month’s topic: OpenAI's GPT-Live Voice AI Architecture and the WARP Protocol.

Chapters

  • (00:00:06) - Intro and Episode Overview
  • (00:01:14) - KlusCon Report: Alberto on Self-Hosted Voice AI Telephony
  • (00:03:19) - KlusCon Report: Hector on eBPF and TURN Servers
  • (00:06:03) - Introducing Varun Singh and Pipecat
  • (00:09:23) - How Voice AI Pipelines Work: Speech to Response Flow
  • (00:12:28) - Turn Detection and Smart Turn Models
  • (00:14:12) - Evolution from Alexa to Modern Voice AI
  • (00:16:49) - Speech-to-Speech vs Cascade Architectures
  • (00:21:06) - The Role of Pipecat in the Voice AI Ecosystem
  • (00:27:50) - Architectural Diagrams: Cascade vs Speech-to-Speech
  • (00:29:30) - Hybrid Fast/Slow Model Loops and Software Engineering Parallels
  • (00:34:19) - Outbound Use Cases and Call Efficiency
  • (00:36:59) - Open Source Speech-to-Speech Models and Education Use Case
  • (00:40:28) - WebRTC vs WebSockets vs Moq for Voice AI
  • (00:47:28) - Build vs Buy: Pipecat and Transport Flexibility
  • (00:50:32) - The Future of Voice-Native UIs
  • (00:55:30) - Wrap-up and Where to Find Varun's Work
  • (00:57:27) - Upcoming Events and Segment Transition
  • (00:58:24) - Analysis: OpenAI's GPT Live and Justin Uberti Interview
  • (01:02:35) - Dual-Path Architecture and Streaming Conversation
  • (01:07:03) - WARP Protocol: Reducing WebRTC Round Trips
  • (01:12:28) - Closing Thoughts and Where to Find More
]]> full 116 false Human speech is often unstructured and messy. To keep up with real conversation, Voice AI bots must parse partial sentences, filler words, and interruptions, all without the visual cues a human listener relies on. On top of that linguistic challenge, real-world latency budgets and network conditions test every Voice AI pipeline, and users don’t forgive one that stutters.

In this episode of WebRTC Live, host Arin Sime talks with Dr. Varun Singh, Chief Product Technology Officer at Daily.co / Pipecat.ai, about the engineering decisions that separate voice agents that succeed from ones that get abandoned. Topics include where Voice AI is working today and where it isn’t, from well-defined tasks like collections to the open-ended conversations everyone is still chasing, and how Pipecat approaches the problem: composing voice agents from STT, LLM, TTS, and everything in between, fitting in speech-to-speech models, and getting from prototype to production. Also discussed: why WebRTC remains the right transport for real-world networks, the production realities of observability, cost, and failure modes, how to weigh build-vs-buy, and where voice-native UIs are headed.

The episode includes our monthly WebRTC Industry Update with Tsahi Levent-Levi. This month’s topic: OpenAI's GPT-Live Voice AI Architecture and the WARP Protocol.

]]> 01:14:03 <![CDATA[Using Voice AI and Avatars to Scale Live Training]]> Wed, 22 Jul 2026 20:51:00 +0000 Arin Sime https://permalink.castos.com/podcast/72791/episode/2670151 https://webrtc.ventures/2026/07/watch-webrtc-live-115-voice-ai-and-avatars-scale-live-training/ How can AI scale clinical roleplay training past what live sessions with human trainers can deliver? CETA Global, a nonprofit that trains frontline mental health providers worldwide, built an AI “flight simulator.” On it, psychologists practice difficult client sessions with an AI patient shown as a lip-synced streaming avatar, get live coaching, and replay feedback afterward. Six AI agents work together to play the patient, coach the trainee, and evaluate the session against evidence-based criteria in real time.

Host Arin Sime talks with Andrés Rincon, the WebRTC.ventures lead developer on EBT-Sim. The team took it from concept to working platform in eight weeks for a Google.org Accelerator: Generative AI demo.

They cover how the team made concurrent agents feel like one conversation: a WebSocket-based real-time layer, Gemini agents orchestrated through Google ADK, low-latency speech-to-text and text-to-speech, and  streaming avatars. Andrés explains why the team chose push-to-talk over automatic turn detection and how they built adaptive resistance into the AI client.

The episode includes our monthly WebRTC Industry Update with Tsahi Levent-Levi. This month’s topic: What's the difference between STUN, TURN & ICE servers?.

Chapters

  • (00:00:18) - Introduction and newsletter mention
  • (00:02:34) - Introducing Andres Rincon and CETA Global's EBT Sim
  • (00:04:11) - What CETA does and why EBT Sim exists
  • (00:07:51) - Why visual avatars matter for training
  • (00:09:52) - Scaling training globally with AI vs humans
  • (00:11:52) - Live demo of the AI flight simulator
  • (00:23:29) - Whiteboard technical architecture deep dive
  • (00:27:18) - Modeling difficult vs easy AI patients
  • (00:30:58) - System architecture walkthrough
  • (00:37:45) - Why WebSockets instead of WebRTC
  • (00:43:26) - Multi-agent coordination with Google ADK
  • (00:47:21) - Lessons learned on streaming avatars
  • (00:51:30) - Closing thoughts and lessons for teams
  • (00:54:27) - Episode wrap-up and announcements
  • (00:55:52) - Meta's AV1 codec blog post analysis
  • (01:01:13) - Asymmetric codec design and device constraints
  • (01:04:19) - AV1 in group calls and future adoption
  • (01:07:02) - Takeaways for smaller WebRTC deployments
  • (01:09:37) - Closing remarks
]]> full 115 false How can AI scale clinical roleplay training past what live sessions with human trainers can deliver? CETA Global, a nonprofit that trains frontline mental health providers worldwide, built an AI “flight simulator.” On it, psychologists practice difficult client sessions with an AI patient shown as a lip-synced streaming avatar, get live coaching, and replay feedback afterward. Six AI agents work together to play the patient, coach the trainee, and evaluate the session against evidence-based criteria in real time.

Host Arin Sime talks with Andrés Rincon, the WebRTC.ventures lead developer on EBT-Sim. The team took it from concept to working platform in eight weeks for a Google.org Accelerator: Generative AI demo.

They cover how the team made concurrent agents feel like one conversation: a WebSocket-based real-time layer, Gemini agents orchestrated through Google ADK, low-latency speech-to-text and text-to-speech, and  streaming avatars. Andrés explains why the team chose push-to-talk over automatic turn detection and how they built adaptive resistance into the AI client.

The episode includes our monthly WebRTC Industry Update with Tsahi Levent-Levi. This month’s topic: What's the difference between STUN, TURN & ICE servers?.

]]> 01:10:18 <![CDATA[An Insider’s Look at OpenAI’s Rearchitected WebRTC Stack]]> Wed, 24 Jun 2026 19:37:00 +0000 Arin Sime https://permalink.castos.com/podcast/72791/episode/2679888 https://webrtc.ventures/2026/06/watch-webrtc-live-114-openai-rearchitected-webrtc-stack/ “Real-time voice AI only works when infrastructure makes latency feel invisible. For OpenAI, that meant changing the shape of our WebRTC deployment without changing what clients expect from WebRTC itself.”

Yi Zhang (LinkedIn), co-author of How OpenAI delivers low-latency voice AI at scale, explains how his team cut pauses, interruptions, and delayed barge-in for ChatGPT voice users and voice agents. He covers running WebRTC inside Kubernetes, separating stateless packet forwarding from stateful session ownership, using ICE ufrag routing to send the first packet to the right server, and what this architecture means for developers building on the OpenAI Realtime API. Yi previously worked at Meta and Microsoft and holds a PhD in Engineering from the University of Wisconsin–Madison.

Also in this episode: our monthly WebRTC Industry Analysis with Tsahi Levent-Levi on where AI tooling fits in WebRTC development.

Chapters

  • (00:00:10) - Introduction and Episode Overview
  • (00:02:42) - Welcoming Yi John from OpenAI
  • (00:03:40) - Yi's Background in Real-Time Media
  • (00:05:05) - OpenAI's Real-Time Infrastructure Team
  • (00:07:23) - Why WebRTC Over WebSockets
  • (00:11:03) - Team's WebRTC Expertise and Approach
  • (00:14:40) - SFU vs Transceiver Architecture
  • (00:22:32) - Transceiver Model Flexibility for Multi-Party
  • (00:27:59) - UDP Port Exhaustion Problem
  • (00:31:28) - Custom Proxy vs TURN/STUNner
  • (00:35:05) - Scaling Stateful Voice Sessions
  • (00:40:16) - Relay and Transceiver Architecture Diagram Walkthrough
  • (00:43:14) - Testing and Performance Tools
  • (00:46:58) - Redis for Session Migration
  • (00:51:13) - ICE Restart and Routing with UFRAG
  • (00:55:45) - Future Direction: Low-Latency Natural Conversations
  • (00:59:31) - Closing Remarks and Upcoming Episodes
  • (01:02:05) - Industry Analysis: Vibe Coding with WebRTC
  • (01:05:14) - Personal Experience Vibe Coding a WordPress Site
  • (01:09:36) - Expertise Still Matters with AI Tools
  • (01:14:36) - AI as a Powerful but Non-Standalone Tool
  • (01:17:34) - Closing of Industry Analysis Segment
]]> full 114 false “Real-time voice AI only works when infrastructure makes latency feel invisible. For OpenAI, that meant changing the shape of our WebRTC deployment without changing what clients expect from WebRTC itself.”

Yi Zhang (LinkedIn), co-author of How OpenAI delivers low-latency voice AI at scale, explains how his team cut pauses, interruptions, and delayed barge-in for ChatGPT voice users and voice agents. He covers running WebRTC inside Kubernetes, separating stateless packet forwarding from stateful session ownership, using ICE ufrag routing to send the first packet to the right server, and what this architecture means for developers building on the OpenAI Realtime API. Yi previously worked at Meta and Microsoft and holds a PhD in Engineering from the University of Wisconsin–Madison.

Also in this episode: our monthly WebRTC Industry Analysis with Tsahi Levent-Levi on where AI tooling fits in WebRTC development.

]]> 01:18:17 <![CDATA[WhatsApp Business Calling and SIP]]> Wed, 13 May 2026 23:45:00 +0000 Arin Sime https://permalink.castos.com/podcast/72791/episode/2680206 https://webrtc.ventures/2026/05/watch-webrtc-live-113-whatsapp-business-calling-and-sip/ WhatsApp has over two billion active users, but for developers building real-time communications applications, it’s long been a walled garden. That’s changing. WhatsApp Business Calling is opening new doors for SIP-based integrations, and the technical landscape around it is evolving fast.

In this episode, Dan Jenkins joins us to break down what combining WhatsApp Business Calling with WebRTC and SIP actually looks like in practice: the use cases, the architecture decisions, and the technical challenges that come with it. Dan Jenkins is founder and CEO of NimbleApe, Everycast Labs, and the organizer of the CommCon conference series. As always, Dan will bring his expertise in low latency media, video streaming and broadcast technologies to our conversation.

The episode include our monthly WebRTC Industry Update with Tsahi Levent-Levi. This month's topic: OpenAI’s Secret WebRTC Architecture: No SFU? No Problem!

Chapters

  • (00:00:00) - Intro and Web Summit Vancouver Highlights
  • (00:02:43) - Digital Samba Interview on Expo Floor
  • (00:04:40) - Welcome Back Dan Jenkins & ComCon Dusseldorf Preview
  • (00:08:54) - What is WhatsApp Business Calling?
  • (00:12:14) - WhatsApp for Healthcare in the Global South
  • (00:16:18) - WhatsApp Client Features and Business Value
  • (00:17:31) - Technical Overview: Setting Up WhatsApp Business Calling
  • (00:21:36) - Business Verification and Phone Number Challenges
  • (00:30:32) - Sandbox Accounts and Production Setup
  • (00:31:42) - Call Flow, Permissions, and Messaging Windows
  • (00:34:49) - SIP Infrastructure Integration Details
  • (00:39:00) - Codecs, Webhooks, and Platform Support (Jambonz, Janus, PipeCat)
  • (00:42:31) - AI Context Continuity Across Messaging and Calls
  • (00:44:59) - Wrap-up: Blog Post Updates and Open Source Benefits
  • (00:51:24) - WhatsApp vs RCS Comparison
  • (00:54:13) - Closing Thoughts and ComCon Plug
  • (00:55:50) - Industry Analysis: OpenAI's WebRTC Scaling Blog Post
  • (00:58:36) - Technical Deep Dive: No SFU, Single Port, Turn Decisions
  • (01:04:20) - Router Architecture and User-Space Optimization
  • (01:06:27) - WebRTC vs WebTransport Future Outlook
  • (01:09:12) - Closing Remarks and Upcoming Events
]]> full 113 false WhatsApp has over two billion active users, but for developers building real-time communications applications, it’s long been a walled garden. That’s changing. WhatsApp Business Calling is opening new doors for SIP-based integrations, and the technical landscape around it is evolving fast.

In this episode, Dan Jenkins joins us to break down what combining WhatsApp Business Calling with WebRTC and SIP actually looks like in practice: the use cases, the architecture decisions, and the technical challenges that come with it. Dan Jenkins is founder and CEO of NimbleApe, Everycast Labs, and the organizer of the CommCon conference series. As always, Dan will bring his expertise in low latency media, video streaming and broadcast technologies to our conversation.

The episode include our monthly WebRTC Industry Update with Tsahi Levent-Levi. This month's topic: OpenAI’s Secret WebRTC Architecture: No SFU? No Problem!

]]> 01:10:35 <![CDATA[Improving End-to-End Quality with WebRTC Observability]]> Wed, 25 Mar 2026 20:05:00 +0000 Arin Sime https://permalink.castos.com/podcast/72791/episode/2689727 https://webrtc.ventures/2026/03/watch-webrtc-live-111-improving-end-to-end-quality-with-webrtc-observability/ Half of the call-quality tickets Balázs Kreith investigates trace back to connectivity problems before media quality ever comes into play. A one-hour meeting with five bad minutes at the start can leave users satisfied, while an aggregated quality-of-service score labels the same call as bad.

Balázs explains how QoS and QoE diverge, why a single score misrepresents what users perceive over the course of a call, and why observability needs to be built in from the start so teams can trace problems after deployment. He is a Senior Software Engineer at Riverside.fm, lead developer of the open source ObserveRTC project, and previously worked on WebRTC teams at Whereby and callstats.io.

Also in this episode: Arin Sime and Tsahi Levent-Levi’s monthly WebRTC Industry Chat on the rise of the WebRTC micro app.

Chapters

  • (00:00:10) - Intro and Auto Scaling Blog Post
  • (00:01:42) - Welcome to WebRTC Live and Guest Introduction
  • (00:03:05) - Why Observability Matters from the Start
  • (00:06:21) - RTC Observability vs Standard Web Analytics
  • (00:08:10) - Balazs's Career Path into RTC Observability
  • (00:13:23) - Understanding WebRTC Stats API
  • (00:16:42) - Quality of Service vs Quality of Experience
  • (00:20:07) - Diagram: Mapping QoS to QoE
  • (00:24:25) - Application Context and Behavioral Indicators
  • (00:26:42) - Connectivity Issues and Real World Debugging
  • (00:32:03) - The Challenge of a Single Quality Score
  • (00:36:53) - Momentum and Time-Series Quality Perception
  • (00:38:46) - Building Feedback Loops with Observability
  • (00:43:21) - Plugging Observe RTC Open Source Project
  • (00:45:21) - Closing Thoughts and Farewell
  • (00:46:31) - Industry Analysis Segment Intro
  • (00:47:18) - Resurgence of Feature-Focused Micro Apps
  • (00:50:46) - AI's Role in Making WebRTC More Accessible
  • (00:53:18) - AI Slop and the Value of Expert Knowledge
  • (00:54:44) - Future of Micro Apps vs SaaS Consolidation
  • (00:58:34) - Closing Remarks and Sign Off
]]> full 111 false Half of the call-quality tickets Balázs Kreith investigates trace back to connectivity problems before media quality ever comes into play. A one-hour meeting with five bad minutes at the start can leave users satisfied, while an aggregated quality-of-service score labels the same call as bad.

Balázs explains how QoS and QoE diverge, why a single score misrepresents what users perceive over the course of a call, and why observability needs to be built in from the start so teams can trace problems after deployment. He is a Senior Software Engineer at Riverside.fm, lead developer of the open source ObserveRTC project, and previously worked on WebRTC teams at Whereby and callstats.io.

Also in this episode: Arin Sime and Tsahi Levent-Levi’s monthly WebRTC Industry Chat on the rise of the WebRTC micro app.

]]> 00:59:21