Block's full-duplex, zero-text Dessa voice AI demos, Mac M4 Max checker & live testing
Dessa Voice AI is an interactive hub showcasing Block's breakthrough full-duplex, zero-text speech-to-speech model. Processing raw audio-in to audio-out without text intermediates, Dessa captures natural turn-taking, interruptions, laughter, and backchanneling. Users can listen to 6 synchronized two-track official demos, run the Mac M4 Max Readiness Checker to verify Apple silicon unified memory compatibility for local inference, and practice real-time voice conversations in the browser. Built for AI developers, researchers, and audio engineers, it also features architectural benchmarks comparing Block Dessa against Sesame CSM, OpenAI GPT-Live-1, and Kyutai Moshi—all free with no login required.
-
Six official synchronized two-track full-duplex voice AI demos
-
Zero-text end-to-end audio-in to audio-out architecture breakdown
-
Mac M4 Max & Apple silicon local hardware readiness checker
-
Live browser-based voice AI interview and conversation testing
-
Side-by-side benchmarks vs. Sesame CSM, GPT-Live-1, and Kyutai Moshi
-
Evaluating full-duplex speech-to-speech voice AI naturalness and latency
-
Checking Mac M4 Max unified memory compatibility for local AI models
-
Practicing live voice interviews and conversational AI interactions
-
Benchmarking zero-text audio models against traditional ASR-LLM-TTS pipelines
-
Demonstrating expressive, interruption-aware voice AI to stakeholders
Shopify Apps
Wix Apps
Ecwid Apps
BigCommerce Apps
