Direct answer: The natural feel of Sesame-style voice comes from keeping end-to-end latency under about 800ms, supporting barge-in (interrupting mid-sentence), and handling turn boundaries with a model rather than a fixed silence timer — get those three right and the interface reads as human.

What’s the latency budget?

Conversational speech tolerates roughly 200–800ms of response delay before it feels off. Past a second, people start talking over the assistant or assuming it didn’t hear them. That budget has to cover speech recognition, the language model, and text-to-speech combined — which is why a naive pipeline (transcribe, then generate, then synthesize, each waiting on the last) blows past it every time.

How do I cut the delay?

• Stream aggressively: start TTS on the first generated clause, not the full reply.
• Use a smaller model for the conversational layer; reserve the big one for hard questions.
• Run recognition and generation on GPU and keep them warm.
• Cap generated reply length — long answers feel slow even when they’re fast.

Streaming is the single biggest win. Waiting for a complete paragraph before speaking is what makes most assistants feel robotic.

How does barge-in actually work?

You need voice-activity detection running during playback, and a path to cancel the current synthesis the moment the user speaks. That means the audio pipeline must be interruptible, not a blocking call that plays to its end. Without cancellation, the assistant talks over the human — the fastest way to make a voice app feel broken.

Fixed timeout or learned turn-taking?

Fixed silence timers are the lazy default: wait 700ms of quiet, then respond. They fail both ways — cutting off a person who paused to think, and waiting forever when someone stops mid-sentence. Sesame’s approach signals the better path: model the turn boundary, using prosody and partial words to predict whether the speaker is done. It’s more work and a genuine upgrade in feel.

What about the voice itself?

Natural prosody matters more than near-perfect pronunciation. A voice that breathes, varies pace, and puts pauses in sensible places reads as real even with an occasional flaw. Flat but flawless TTS lands squarely in uncanny valley.

What should I prototype first?

Build the interruption path and measure latency end-to-end before touching prompt quality. If you can’t get under a second and can’t be interrupted, nothing else matters. Once those hold, tune the conversation.

Key Takeaways

• Aim for 200–800ms end-to-end latency; past 1s users assume it failed.
• Stream TTS from the first clause instead of waiting for a full reply.
• Barge-in requires cancellable audio, not blocking playback.
• Learned turn-taking beats fixed silence timers on both failure modes.
• Prototype latency and interruption before polishing prompts or voices.

Direct answer: Sesame AI is a voice conversation model accessible through a web demo where you talk to a persona like Maya or Miles; it focuses on natural, low-latency speech rather than task execution, and it’s free to demo in supported browsers.

How do I start a conversation?

1. Open Sesame’s demo page in Chrome or Safari.
2. Allow microphone access when prompted.
3. Pick a voice persona (Maya or Miles).
4. Just start talking — hold a normal conversation and let it respond.

Headphones with a mic help a lot. Speaker feedback causes the model to talk over itself, which makes the chat feel less natural than it is.

What does the voice actually feel like?

Warmer and more continuous than most assistants. It handles interruptions better than turn-based systems, so you can talk over it and it adapts instead of stopping dead. The pauses land in more natural places, which is a small thing that changes how the whole conversation feels. It’s still an AI, but the uncanny gaps are narrower than what you get from a standard voice assistant.

What can I use it for?

Conversation practice, thinking out loud, and getting a sense of where voice interfaces are heading. It’s not a productivity tool — it won’t book your calendar or write your code. Treat the demo as a preview of a voice layer that other apps might put on top of a task-performing model.

Is Sesame free?

The demo is free to use in a supported browser. Sesame’s business is the underlying voice technology, so the public demo is a showcase rather than the product itself. There’s no paid consumer tier to compare against yet — which is worth knowing before you plan workflows around it.

Where does it fall short?

It’s a demo, and the edges show. Conversations can lose the thread over long sessions, and it’s weaker on factual recall than text-first assistants — it sounds confident while being wrong, which is a bad combo. Browser and mic setup matters; a bad microphone ruins the effect. And there’s no persistent memory across sessions, so each call starts from scratch.

Who is it for?

Anyone curious about voice AI’s current ceiling, and builders evaluating a speech layer to embed elsewhere. If you need task execution, this isn’t it yet.

Key Takeaways

• Sesame AI is a voice-first assistant with personas Maya and Miles, demoed in-browser.
• The demo is free; Sesame’s real product is the underlying voice tech.
• Handles interruptions and natural pauses better than turn-based assistants.
• No persistent memory between sessions and weaker factual recall.
• Use it to explore voice interfaces, not to run tasks.