Topic
All digests tagged Voice AI
Suggested route
Move from earlier context through the available material to the newest signals. This route is a reading aid, not a difficulty ranking.
AssemblyAI introduces TTS-1, a frontier TTS model built for conversational voice agents. By owning the entire voice loop (STT to TTS) on a unified inference stack, they offer scalable, low-cost voice AI with uncapped rate limits.
Read article
The session explores the future of AI, from open models to physical AI. Key topics include digital twin simulation, agentic orchestration, and multi-modal generative models like Flux 3 for robotics and manufacturing.
Read article
Jesse Hall explains that for voice agents, latency is a budget, but the goal is human-like conversation. He details why component leaderboards fail and how to build a robust, real-time pipeline.
Read article
Learn how to build domain-specific voice agents by fine-tuning ASR and LLMs. We cover low-latency architecture, WebRTC best practices, and advanced model customization techniques.
Read article
Building a Voice OS requires moving beyond accurate ASR to mastering latency, personalization, and style inference. The talk details using LLMs and RL to structure raw speech into polished, usable communication.
Read article
Learn how AssemblyAI built 'Joey,' an AI agent that boosted ticket resolution from 10% to 80%. Discover the architecture using local markdown, embeddings, and the Claude Agent SDK.
Read article
Unmute is an open declarative standard for voice agents that solves vendor lock-in and prompt fragility. By compiling agent behavior into portable packages, it ensures structural guarantees and portability across platforms like LiveKit and PipeChat.
Read article
The Gemini Live API introduces async function calling, Proactive Audio, and context injection via sendClientContent. The demo showcases frontier-level background reasoning and model switching for superior creative output.
Read article
Improving voice AI requires optimizing system behavior under uncertainty, not just accuracy. By minimizing the Outcome User Cost Heuristic (OUCH), developers can significantly boost UX even if the core recognition accuracy remains constant.
Read article
Voice AI failures are not random bugs, but structural issues. Kim introduces a linguistic framework mapping communication to two channels (listening/speaking) and four interdependent levels: sounds, words, interaction, and mental model.
Read article
Voice AI agents often fail in production due to latency, brittle data collection, and poor LLM-to-TTS handoffs. The talk outlines solutions using structured data modeling and normalization layers.
Read article
Building a voice-first AI companion requires designing for 'conversational volatility.' The team uses tiered model routing based on emotional stakes and rebuilds context per turn, treating memory as a retrieval system.
Read article
The talk details the evolution of voice agents, arguing that achieving full-duplex, natural conversation requires a hybrid architecture: coupling a small S2S interface with a powerful background text LLM.
Read article
Building smarter voice agents requires managing complex challenges like subsecond latency, multi-party turn-taking, and user trust. The discussion contrasts structured healthcare flows with open networking interactions.
Read article
Gemini 3.5 Transcribe is an LLM-based transcription model available via Interactions API and Live API, enabling fast, contextually accurate transcribing of multi-speaker recordings.
Read article
Building reliable voice agents requires complex cascaded pipelines (STT $\rightarrow$ LLM $\rightarrow$ TTS). Engineers must manage latency trade-offs, context limits, and model fallbacks to achieve human-like performance.
Read article
An autonomous AI agent handles product launch positioning (Tempo) while AFK, escalating only critical decisions via a voice call and logging all outcomes directly into project files.
Read article
Watch an AI coding agent autonomously call a developer to resolve a critical checkout API bug. Learn how Voice AI and Claude Code CLI manage complex technical trade-offs (Option A vs B) in real time.
Read article
Building production voice agents requires a modular, cascading architecture (ASR $\rightarrow$ LLM $\rightarrow$ TTS). Key focus areas include managing latency within a 1-1.5 second budget and implementing robust fallback systems across all components.
Read article
Learn how to boost voice agent transcription accuracy with Agent Context Carryover. This feature automatically feeds conversational context into LiveKit's Universal 3.5 Pro model, simplifying complex AI integrations.
Read article