Home▸AI & Machine Learning▸Voice Assistant Endpointing & Barge-In Lab

Voice Assistant Endpointing & Barge-In Lab (2D)

Interactive 2D canvas simulator of a voice-assistant conversation pipeline: watch a real voice-activity detector, silence-timeout endpointing and barge-in interruption logic decide, frame by frame, when the assistant should reply.

AI & Machine Learning2DModerate60 FPS📱 Mobile-adapted⇄ 3D version

Details

CategoryAI & Machine Learning

RenderingCanvas 2D

DifficultyModerate

Target FPS60

ControlsDrag the energy-threshold slider to tune how loud speech must be to trip the voice activity detector, the endpoint silence-delay slider to set how long a pause must last before the assistant replies, the playback-speed slider to speed up or slow down the simulated conversation, and toggle barge-in on/off to compare full-duplex and half-duplex turn-taking.

voice AIconversational AIVADendpointingturn-takingbarge-in

Related simulations

Conversational AI: Chatbot Dialog Manager & RAG SimulatorInteractive 3D simulation of a chatbot's architecture: a dialog manager holding the…

Social AI: Turn-Taking PredictorA 3D social-signal-processing simulator where an AI model scores every listener on…

Eval Pipeline Cost Landscape: Batching, Caching & Shadow Tests (2D)2D parameter-space heatmap of an LLM evaluation pipeline: sweep batch size against cache-hit… ⊞ All 1161 in AI & Machine Learning →

Before a conversational-AI system can generate a reply, it has to solve a much older and stricter problem: knowing exactly when the human stopped talking. This simulator renders that decision loop as two scrolling 2D waveform lanes — a live microphone track and the assistant's own speech track — driven by a real short-term-energy voice activity detector with exponential smoothing and hangover, a silence-timeout endpointing rule that decides when to hand the turn to the assistant, and a barge-in monitor that can cancel the assistant's speech mid-sentence the moment the user starts talking again. Drag the energy threshold and endpoint delay to see how a real voice assistant balances snappy turn-taking against cutting people off, and toggle barge-in to compare full-duplex and half-duplex conversation design.

⚙ Under the hood

A real-time 2D canvas simulator of a voice-assistant conversation pipeline: an energy-based voice activity detector with hangover, a silence-timeout endpointing rule, and full-duplex barge-in interruption logic, all driving two scrolling waveform lanes.

voice AIconversational AIVADendpointingturn-takingbarge-in

2D · HTML5 Canvas 2D · 60 FPS target · runs fully client-side, no install