News

This is Sesame: The ‘human voice’ generated by AI

Sesame's AI-generated voices, Maya and Miles, replicate human speech with striking realism, raising new possibilities—and concerns—about our emotional connection to machines.

This is Sesame: The ‘human voice’ generated by AI
Agencias

Agencias

  • March 6, 2025
  • Updated: March 6, 2025 at 7:55 AM

Artificial intelligence is advancing at an astonishing pace, and Sesame’s latest voice model is pushing the boundaries of human-like speech. While chatbots like ChatGPT have long allowed us to converse with AI through text, Sesame introduces something even more immersive: a voice interaction so natural that it blurs the line between human and machine.

A voice that feels real

Sesame’s technology relies on a Conversational Speech Model (CSM), which replicates the nuances of human speech with incredible accuracy. Unlike traditional text-to-speech systems, this model integrates pauses, intonations, and emotional subtleties, creating a conversational experience that feels astonishingly real.

Users who have interacted with Maya and Miles, the AI-generated voices from Sesame, report feeling an emotional connection—an outcome rarely seen with previous AI-generated speech. Some even describe the experience as “strange, exciting, and unsettling all at once.” The realism is so striking that it raises new ethical and psychological questions about human relationships with AI.

The secret to Sesame’s success lies in a dual-model architecture based on Meta’s Llama framework, consisting of a primary AI engine and a specialized decoder. This combination enables rapid response generation without noticeable latency, ensuring fluid and dynamic conversations. The company has trained these models using one million hours of English-language audio, refining speech patterns to near-human perfection.

Despite its impressive capabilities, Sesame’s AI is still imperfect. Users note occasional unnatural responses, awkward prosody, and inconsistencies in conversational rhythm. The company acknowledges these limitations but remains confident in its ability to refine the technology further.

As AI-driven voices become more advanced, the future of human-AI interaction is shifting dramatically. Whether this will enhance our lives or introduce new challenges remains to be seen, but one thing is certain—Sesame has taken us one step closer to an AI-powered reality.

Latest Articles

Loading next article