Dessa AI Voice Demos

Talking about the release

Listen for the timing of laughter and changes in volume. Block describes the recording as a live exchange in which Dessa chooses its response and delivery. Play the original recording below without leaving this page.

Remembering a movie

This example is useful for listening to sentence completion and conversational timing. Selected examples can illustrate a behavior without establishing how consistently it occurs.

An anniversary surprise

Block presents this clip as an example of an emotional-sounding reaction. Expressive audio is a generated behavior; it does not establish that a system experiences emotion.

Life advice and expressive conversation

Listen to pace, pauses, and intonation as well as the words. Consider whether the model tracks context and whether interruptions actually change what it says.

Expressive conversation

A conversation about coffee, laughter and stress. Listen for sentence completion, a joke and changes in delivery. Use the highlight buttons to compare short reactions with longer responses.

Dessa in conversation with GPT-Live-1

A model-to-model exchange for listening to interruptions, patience and short acknowledgments. Both voices are generated; this is a selected recording rather than a benchmark.

Dessa and GPT-Live-1 are AI-generated voices. The noisy recording also includes a human. No endorsement by the third-party model developer is implied.

More conversation samples

Explore 15 additional full recordings below. The titles identify the topics in the developer’s collection; they do not establish tested abilities or multilingual support.