Talking with your bot

Voice turns your chat into a spoken conversation. Hold the mic and talk instead of typing; flip on read-aloud and your bot's replies come back as audio, with the words highlighted as they're spoken.

It isn't a separate app or a different bot — it's the same chat you already have, with a mic and a speaker added to it.

Speaking instead of typing

On desktop, hold the mic button in the chat (or hold Alt) to record — release and your speech becomes the message you'd otherwise have typed. On mobile, a small floating mic sits over the chat; drag it out of the way if it's in your view, and it remembers where you left it.

Your recording is transcribed and turned into your chat message. The audio itself isn't kept — once the words land in the chat, the recording is gone.

Ask: "Just talk it through with me," then hold the mic and describe what you need instead of typing it out.

Hearing replies read aloud

Turn on read-aloud and your bot's answers play back as speech, with each word highlighted as it's spoken — easy to follow along, easy to skim ahead. Play any reply on demand from its message bubble, whether or not read-aloud is on.

Turning it on

Voice lives in Settings → Voice in the dashboard. From there you can turn push-to-talk on or off, choose whether releasing the key sends the message immediately, turn read-aloud on or off, pick the spoken language (Swedish, English, or auto-detect), toggle sound effects, and set a voice gender per bot.

There's nothing to install — voice becomes usable in a chat as soon as a working voice key resolves for that bot, which happens one of three ways:

• Your own key. Connect a personal OpenAI or ElevenLabs API key in Settings, and voice works immediately — no admin or organization action needed. Personal keys are never metered by GATE; your provider bills you directly.
• Your organization's shared key. If your org has connected its own ElevenLabs key and you've chosen to use it (rather than a personal one), your speaking rides on that key instead.
• GATE's built-in allowance. If your organization has been granted GATE's own voice allowance, read-aloud works out of the box on GATE's key, up to a monthly character budget the organization is given.

Limits, honestly

• Read-aloud (text-to-speech) is the metered half. On GATE's built-in allowance, replies are counted against a monthly character budget. Run out, and read-aloud pauses until the budget resets next month — typing and reading replies as text still works fine.
• Speaking to your bot (speech-to-text) isn't capped on any path today — only read-aloud has a budget.
• A key you bring — personal or your organization's — is never metered by GATE. The provider bills that account directly; only GATE's own built-in key carries the allowance.
• If a shared or platform key is temporarily unavailable, GATE falls back gracefully rather than breaking your chat — worst case, read-aloud is unavailable and typed chat continues as normal.

Going deeper

• Built-in tools — the rest of what your bot can do on request
• Credentials & delegation — how personal API keys and other credentials are stored