How to Use Muse Voice: Talk to Muse Live (2026)

Meta switched on real-time voice for Muse on 5 November 2026. You can now talk to Meta’s personal AI agent out loud — and watch an expressive avatar answer you back — instead of typing every request. The feature, called Muse Voice, is rolling out in the Muse app in the US on mobile, the web and Mac. This guide shows you how to start a call, how to pick a voice and avatar, and what it is actually good for.

Quick answer: Muse Voice is live in the US from 5 November 2026. Tap the phone icon, pick a voice, and talk — Muse answers in real time with an expressive avatar.

This is a new release, not the basic voice mode covered in earlier tips pieces: Meta previewed real-time voice at Connect 2026 on 23 September, and 5 November is the date it announced the US rollout. Meta’s chief AI officer, Alexandr Wang, posted a demo call on X the same day, writing that “muse voice is out for everyone in the US!”

What Muse Voice actually is

Muse Voice turns Muse from a text agent into a conversation you can have out loud, with a face that reacts while you talk. Two pieces of technology power it, both from Meta: Muse Realtime Voice, which produces spoken answers immediately, and Muse Realtime Avatar, which generates the animated face. Meta describes them as one streaming system — the same stream of speech tokens drives the voice and the avatar, so the lip movement, expressions and gestures stay synchronised with the words.

Three things make it different from an ordinary voice assistant. First, you can interrupt Muse mid-answer, the way you would cut in on a phone call, and it picks up the new thread. Second, Muse keeps working on tasks in the background while you keep talking — you can hand it a job and discuss the next one at the same time. Third, the avatar is a visual presence, not a waveform: it reacts while you talk, so you can tell it is following along. Meta says the avatar video carries an invisible Meta Video Seal watermark, and that Muse is for users aged 18 and older.

Meta’s research team reports roughly 870 milliseconds from the end of your turn to the first synchronised voice-and-video response, rendered as 448-by-768 portrait video at 25 frames per second. Those are Meta’s own figures, not an independent measurement — treat them as the design target, not a promise about every call.

How to start a Muse Voice call

Starting a call takes one tap — but only if you are in the US and the rollout has reached your account. Meta says voice is rolling out in the US to start, and it has given no date for other countries. Some users reported seeing voice chat from mid-October, during a staged rollout; 5 November is the date Meta announced it for everyone in the US.

Open the Muse app on your phone, on the web, or on a Mac. You need to be signed in to Muse, which is available to US adults aged 18 and over.
Tap the phone icon in the top-right corner of the app. That is the whole entry point — there is no separate voice app and no setting to switch on first.
Start talking. Say what you need the way you would say it to a person. You do not need a polished prompt; Meta’s own framing is that you can “talk through an idea with your Muse and let it take shape without worrying about the exact prompt.”
Interrupt whenever you need to. If Muse is heading the wrong way, cut in mid-sentence and redirect it. You do not have to wait for a response to finish.
Hand it work while you talk. Ask Muse to do something — research a trip, draft a reply, compare options — and keep talking while it works in the background. It returns when it needs approval, for example before a sensitive action such as a purchase.

When you are done, end the call — the conversation stays in your Muse history like any other chat, so you can pick it up in text later.

Pick your voice and avatar

Muse Voice ships with hundreds of voices, and you can change yours whenever you want. Meta illustrates the range with two examples: “the energy of a New Yorker” and “the flair of a pirate”. The practical move is to pick a voice that matches the job — a calm voice for rehearsing a difficult conversation, a brisk one for working through a to-do list — and to change it when the job changes.

The avatar is customisable too. TechCrunch reports that the default avatar goes by the nickname Jolly and can be changed into a person, an animal or another character. The avatar reacts as Muse listens and speaks, using expressions and gestures suited to the tone of the conversation, so a tense exchange looks different from a light one. If the face distracts you, switch to voice-only — the conversation works exactly the same way without it.

Avatar setting	What you see	Best for
Voice only	No avatar, just the conversation	Calls while walking, driving or cooking; long working sessions
Avatar beside the chat	A small animated face next to the text	Checking what Muse is doing while you talk
Full screen	A face-to-face video-style view	Rehearsals, interviews and conversations that feel personal

Five things worth trying on your first call

Voice is best for jobs where thinking out loud beats typing — half-formed ideas, rehearsals and hands-free moments. Meta suggests the same three starting points in its announcement: developing an idea, planning a trip, and rehearsing a job interview or a presentation. Here are five spoken prompts to try first, written the way you would actually say them.

Shape a half-baked idea. “I have a rough idea for a side business — a weekend food stall — but I don’t know where to start. Ask me questions until the idea has a shape.” Let Muse interview you instead of asking it for a finished plan.
Plan a trip from three facts. “I want to see Kyoto, I like waking up somewhere quiet, and there’s one meal I want to plan the whole trip around. Build me a three-day plan and tell me what to book first.” This is Meta’s own example, and it plays to Muse’s strength: it can keep working on the booking research while you talk.
Rehearse out loud. “Pretend you’re interviewing me for a marketing job. Ask me one question at a time, then tell me what I did well and what to fix.” Hearing your own answer, and answering in the moment, is the point — a typed rehearsal cannot do this.
Work hands-free. Start a call while you cook or walk, and say: “While I’m busy, find me three options for a birthday gift under $50 and read me the best one.” Muse works in the background; you only need to listen.
Make a quick decision. “I’m choosing between two laptops. Ask me three questions, then recommend one and tell me what I’d regret about the other.” Voice forces a short, honest answer — and the interruption button means you can stop Muse the moment it drifts.

One guardrail applies to all five: Muse still asks for approval before sensitive actions, such as purchases. Saying “book it” does not skip that step — and that is a feature, not a bug. Our earlier coverage of Muse’s shopping carts explains how that approval step works in practice.

Voice or typing: when to talk, when to type

Talk when the idea is unfinished; type when the wording matters. Voice wins for brainstorming, rehearsals and anything you do with your hands busy. Typing wins for anything you need to keep word-for-word — a prompt you will reuse, a message you will send, or a set of instructions with exact figures. If you are not sure, start by talking and switch to text when the plan firms up; the conversation carries over.

Job	Better by voice	Better by typing
Brainstorming an idea	Yes — thinking out loud helps	No
Rehearsing an interview or pitch	Yes — timing and tone matter	No
A prompt you will reuse	No	Yes — exact wording counts
Hands-free tasks (cooking, walking)	Yes	No
Detailed research with figures	Partly — ask, then read the text	Yes — for the numbers
Shopping with approval steps	Yes — talk, then approve in the app	Partly — for the final check

What Muse Voice can’t do yet

The honest limits are availability first, devices second. Voice is rolling out in the US only. Muse itself is still not listed on Canada’s App Store, and Meta has given no date for Canada or for this voice layer there — and none at all for the UK, India or anywhere else. If you are outside the US, there is nothing to switch on and no waitlist Meta has documented; the availability picture we track here still shows the US and Canada as the only launched markets, with Japan and Singapore added in late October.

Voice conversations are also coming to Meta’s AI glasses and to Muse Charm, a pocket-sized device Meta describes as built for speaking and interacting with Muse — but neither is part of this rollout. TechCrunch reports the Charm, a small wearable with a screen for the avatar, is planned before the end of the year. Until then, voice lives in the Muse app on mobile, the web and Mac only — not in WhatsApp, where Muse remains text-only. And one standing caveat: this guide is desk-researched from Meta’s announcement and independent coverage, not hands-on tested, because the rollout is US-only.

Our take

Muse Voice matters less as a voice feature than as a change in how you direct an agent. Typing forces you to write a good prompt before you know what you want; talking lets Muse do the clarifying work with you, in real time, while it is already doing the task in the background. That is the real shift here — the avatar is the visible part, but the background working is the useful part.

The first-call test is simple: take one job you would normally type out, start it by voice, and interrupt Muse twice. If the conversation survives the interruptions and the task still gets done, voice is your faster route for that kind of job. If you find yourself repeating the same instruction, go back to typing — some jobs still need exact words. For reusable wording once you know what works, our copy-paste Muse prompts collection is the typed companion to this guide.

Frequently Asked Questions

Is Muse Voice available in India or the UK?

No. Meta says voice conversations and expressive avatars are rolling out in the Muse app in the US to start, and it has given no date for other countries. Muse itself is available in the US and Canada, with Japan and Singapore added in late October 2026; India has no confirmed launch date.

Does Muse Voice work on WhatsApp?

No. Voice is available in the Muse app on mobile, the web and Mac. On WhatsApp, Muse remains text-only — the announcement covers the app, not the WhatsApp integration.

Can I change the avatar and the voice?

Yes to both. Meta says there are hundreds of voices to choose from and you can change your selection whenever you want. TechCrunch reports the default avatar is nicknamed Jolly and can be changed into a person, an animal or another character, and you can also choose voice-only, avatar beside the chat, or full screen.

Can I interrupt Muse while it is talking?

Yes. Muse Realtime Voice supports interruptions, so you can jump in with a follow-up or a correction rather than waiting for a response to finish. Muse keeps working on background tasks while the conversation continues.

When is Muse Voice coming to Meta’s AI glasses?

Meta says voice conversations are “coming soon” to AI Glasses and Muse Charm, with no date given. TechCrunch reports the Charm, a small wearable with a screen for the avatar, is planned before the end of the year.

Sources and methodology

Claims in this article rest on Meta’s own announcement, “Give Your Muse a Voice” (about.fb.com, 5 November 2026), checked against independent coverage from Unite.AI and iPhone in Canada (both 5 November 2026), and Alexandr Wang’s launch post on X (5 November 2026). Latency and rendering figures are Meta’s own research numbers, labelled as such; the “Jolly” avatar name and the Muse Charm timing come from TechCrunch reporting via TechProlonged, not from Meta. All facts were verified on 6 November 2026. This guide was desk-researched, not hands-on tested, because the rollout is US-only.

Govind Dheda

Govind Dheda covers AI tools, guides and prompts for OpenAIMaster — what is new, what is worth using, and how to put AI to work.