D
DessaAI Voice
Log in
AI-Powered Voice Chat & Text-to-Speech

Dessa AI Voice.
Talk Naturally with AI.

Experience natural AI voice conversations with responsive speech and expressive audio. Use Dessa AI Voice to explore voice chat and text-to-speech in one simple, browser-based experience.

Start a voice conversation or turn text into speech with an easy-to-use AI voice tool. Explore available features and find the right voice experience for your needs.

Try AI Voice Chat Try Text to Speech
100 Dessa Voice Credits
Microphone Ready

Click Start Talking, allow your microphone, and speak naturally to Dessa AI.

Full-Duplex audio pipeline · Voice turns transcribed in real-time · Speech-to-Speech synthesis

Real-time edge speech transformation powered by Dessa AI architecture.

Live Voice Captions Microphone standby

Press "Start Talking", say something into your mic, and watch your words transcribe and Dessa AI Voice respond in real-time.

Dessa AI Voice Model • Audio In, Audio Out • Block Official Research ↗

Dessa AI Voice Live Demo: Remembering a Movie

Listen to an authentic unsteered recording of a person talking to Dessa AI Voice. When the speaker blanks on a title, Dessa AI Voice jumps in, completes the sentence, steers the chat, and shares a genuine mid-sentence laugh.

Human Speaker Human Speaker
Dessa AI Voice
0:00 / 2:39

AI Voice Demos: Emotion, Expression & Conversation

Explore audio examples of expressive speech, conversational nuance, and real-time voice interaction. Demonstrations reflect the specific recordings and systems identified below.

TRACK 01 VOICE EXPRESSION

AI Voice Demo: Laughter & Vocal Expression

Listen for changes in vocal expression, including laughter, emphasis, and shifts in speaking volume. These examples illustrate how vocal delivery can add nuance beyond the words themselves.

TRACK 02 EMOTION & NUANCE

AI Voice Demo: Empathy & Conversational Nuance

This conversation explores how tone, pauses, pitch, and vocal emphasis can shape an emotionally nuanced exchange. Listen to the speakers' delivery and how it contributes to the conversation.

TRACK 03 CONVERSATIONAL TIMING

Voice AI Demo: Interruptions & Turn-Taking

Explore conversational timing, interruptions, pauses, and brief verbal responses such as "yeah" and "got it." These moments illustrate how turn-taking can affect the flow of a voice conversation.

Audio examples are provided for demonstration and educational purposes. Model capabilities vary by system and implementation. Any third-party recordings or model references should be clearly attributed to their original sources.

What Is Dessa AI Voice? Understanding Speech-to-Speech AI

Dessa AI Voice introduces the possibilities of AI-powered voice interaction, where spoken input can be used to generate spoken responses. Unlike traditional cascaded voice systems that process speech through separate speech recognition, language processing, and speech synthesis stages, speech-to-speech approaches can model the relationship between incoming and outgoing audio more directly.

Voice carries more than words. Tone, rhythm, pauses, laughter, and emphasis can all contribute to meaning. How well an AI system preserves or responds to these signals depends on its underlying model, architecture, and implementation.

How Speech-to-Speech AI Differs from Traditional Voice Pipelines

A traditional voice pipeline typically converts speech into text, processes the recognized text with a language model, and then generates spoken audio. Speech-to-speech systems can reduce the need for explicit intermediate text representations, depending on their design.

Consider the sentence "I didn't say she stole my CUDA code." Changing which word you emphasize can change its meaning. This illustrates why prosody and context matter in spoken communication.

Different voice AI systems use different architectures, training methods, and deployment environments. Actual latency, conversational responsiveness, emotional expressiveness, and hardware requirements depend on the specific model and how it is deployed.

Architectural Dimension Cascaded Voice AI Speech-to-Speech AI
Processing Pipeline Speech recognition → Language model → Speech synthesis Incoming audio → Audio-based response generation
Conversational Latency Depends on the models, network, and processing stages Depends on the model architecture and deployment
Prosody and Emotion Depends on recognition, context handling, and voice synthesis May preserve or model acoustic and expressive cues more directly
Deployment Requirements Cloud, local, or hybrid deployment Cloud, local, or hybrid deployment, depending on the model

This comparison describes general voice AI architectures, not a benchmark of specific products. Performance and capabilities vary by implementation.

The Team Behind Dessa AI Voice

Dessa AI Voice was built by a specialized research lab at Block, led by Ragavan Thurairatnam and reporting directly to Jack Dorsey.

RESEARCHER Louay Hazami
RESEARCHER Hashiam Kadhim
RESEARCHER Qi (Alex) Liu
RESEARCHER Rayhane Mama
RESEARCHER Aydin Polat
RESEARCHER Marc Tyndel
INTERN Jibran Iqbal Shah
INTERN Dvir Zagury-Grynbaum

Frequently Asked Questions About Dessa AI Voice

What Is Dessa AI Voice? +

Dessa AI Voice is a voice AI experience focused on natural, real-time spoken interaction. Instead of treating every exchange as a separate text-to-speech task, it aims to make conversations with AI feel more fluid, responsive, and natural. Explore the available features on this website to learn how Dessa AI Voice fits your voice interaction needs.

How Can I Try Dessa AI Voice Chat Online? +

You can get started by opening the Dessa AI Voice Chat experience on this website, if the interactive demo is available. Follow the on-screen instructions to begin a voice session, speak naturally, and explore the available conversational features. Microphone access may be required for voice input.

What Makes Dessa AI Voice Different from Traditional Text to Speech? +

Traditional text to speech primarily converts written text into spoken audio. Dessa AI Voice focuses on conversational voice interaction, where the experience can involve listening, responding, and maintaining a more natural speaking rhythm. The exact capabilities depend on the model and features available in the current implementation.

Does Dessa AI Voice Support Real-Time Voice Conversations? +

Real-time voice interaction is a key use case for Dessa AI Voice. The responsiveness of a conversation depends on the underlying model, audio processing, network conditions, and device performance. Check the live demo or current product documentation for the real-time features supported by this website.

Can I Interrupt Dessa AI Voice During a Conversation? +

Natural conversation often involves interruptions, quick follow-up questions, and changes of direction. Whether you can interrupt Dessa AI Voice while it is speaking depends on the full-duplex capabilities enabled in the current implementation. Try the interactive experience to see how it handles turn-taking and interruptions.

Can Dessa AI Voice Run Locally Without an Internet Connection? +

Local inference and offline operation depend on whether the underlying model weights and runtime are available for local deployment. Hardware requirements can also vary by model size and configuration. Check the official model documentation to confirm local availability, supported devices, performance requirements, and whether an internet connection is needed.

Is Dessa AI Voice Free to Use? +

Availability and pricing depend on the features offered by this website. Check the current demo, pricing page, or usage terms to see whether Dessa AI Voice is free to try and whether any usage limits, account requirements, or paid plans apply.