Rooms, and how they start

A room never launches empty or unprepared. Every room passes through the same warm-up before a single voice goes wide — then it runs in one of two formats, picked by the host when they open it.

Start with your key points

Before a room can open at all, you choose one of your idea collections — a saved title plus a handful of sharp key points (at least three). Those points become the room: its title, its hashtags, and the little map pinned to the center of the stage once you're live. They do two jobs at once — a gentle teleprompter so you never blank mid-room, and a shared map so your listeners can see where the idea is going.

If you can't boil your take down to a few clear points, you don't have a take yet — you have a feeling. Writing the points is the first rep.

Idea collections are reusable: build one once and launch a room from it whenever the urge strikes. (They also power the deck in a Live Class.) See Getting started for how to make one.

The backstage warm-up

Every room initializes in a hidden, zero-pressure backstage state. Nobody outside can see it, and — this is the important part — nothing you say is broadcast yet. Your microphone track stays local to your device; no audio leaves your machine. It's a private mirror to rehearse your points out loud before anyone arrives. A soundcheck, not a performance.

The room stays backstage until a real little audience has gathered:

• Exactly 1 Sparker (you)
• At least 1 Miner
• At least 1 Lurker

The instant that's met, the room flips live for everyone at once — your local track goes wide and the room's tools unlock. There's no countdown and no round clock: a live room simply runs until you end it. If everyone drifts off, the room tidies itself up, closing about a minute after the last person leaves.

That Lurker in the gate isn't a formality. A room with speakers but no crowd has nobody to react — and the reactions are half of why anyone shows up. Waiting for that first Miner and Lurker to wander in is exactly the gap the Waiting Hall closes: instead of hoping, you pull available people straight into the seats you need.

Normal rooms

The real-time format — pure live voice. You drop your take in English; the two Miners think out loud with you, asking and pushing for the clearer version; the crowd listens in and feeds tips from the side. It's built for confident speakers who want the genuine article, with no translation bottleneck and no waiting.

Under the hood, voice is full-duplex among the Sparker and the two Miners, so the three of you can talk and even gently interrupt naturally. Three is also about the most voices that can share a clean, lag-free live mix peer-to-peer — another quiet reason the stage caps at one Sparker and two Miners. That live mix is then streamed one-way to every Lurker with low latency, so an unlimited silent crowd can listen in without straining anyone's connection. A few things define the feel:

• Webcam is encouraged for the active speakers — a bit of eye contact keeps it human and accountable.
• Your key points sit in the center as a shared map the whole time.
• If a Miner is spent, Request a Sub hands their seat to a waiting Lurker without a gap.

For the live tools once you're in, see In the room.

Audio Chat rooms

The gentle format — built for time-abundance and zero stage fright. This is the only room with an Echoer, an interpreter attached to the Sparker's drafting flow. Instead of a continuous live stream, the stage is a clean timeline of dubbed cards: short messages that carry both text and voice. Nothing here is ever live — every card is quietly built off-stage before it appears.

A Sparker's card, when there's an Echoer, is a two-person build:

1. Draft. The beginner Sparker types a raw take — Vietnamese or rough English — into a private draft only the Echoer sees.
2. Polish. The Echoer rewrites it into natural, idiomatic English and hands it back.
3. Two dubs. Both record against the polished text: the Sparker dubs the line in their own voice — their real attempt — and the Echoer records their own short interpretation, "what they mean is…"
4. Send, together. The card only drops onto the timeline once both have submitted. It then carries the text plus both recordings, and anyone can play either voice.

Miners — and a Sparker flying solo without an Echoer — use a lighter self-dub: type, record, send, one voice, no hand-off. And if the words won't come at all, blank voice lets you ship a card as text only. The goal is to keep the timeline moving at whatever level of fluency you've got that day.

Voice here is private between the Sparker and the Echoer while they draft; what reaches the room is the finished card. The result removes the two scariest parts of speaking — "I don't know how to say this" and "I have to say it right now" — leaving a calm, manageable challenge.

And because every card is a clean block of text plus audio, the whole conversation can be saved for later — download the room's exchange as a tidy archive and keep the lines you wrote, polished, and voiced, long after the room has closed.

Choosing a format