Blog/Published: Updated:

Taking Part in a Meeting You Cannot Hear

Gaavala lets you follow a browser meeting through live translated captions and answer with typed text that a synthetic voice plays into the call. Typed replies require Pro and a supported meeting microphone connection. This guide explains the setup, platform differences and limits of playback confirmation.

Original + translation, paired
Timestamps, language badges, and both sides of every sentence — what was said and what it means stay together.

The reading half

This is what Gaavala already does for everyone, and it is the same machinery a hearing user gets for translation.

The extension captures the meeting tab's audio and streams it to a speech-to-text engine, which returns a transcript as people speak. That transcript lands in two places: the side panel, as a scrolling bilingual record with timestamps and language badges, and a floating layer over the meeting itself, so you are not looking away from faces to read words.

Two details matter more for a deaf reader than for a translation reader:

Speakers are separated and coloured. Gaavala detects who is talking and assigns each voice a consistent colour across the overlay, the side panel and your exports. You can rename "Speaker 2" to a real name mid-meeting and every surface updates. In a fast four-way exchange this is the difference between a conversation and a wall of text.

Newer lines are visually stronger. In the overlay, fresh lines land at full strength and older ones fade back, so your eye finds the current sentence without hunting.

Captions are not Pro-gated by feature — the free allowance is five minutes, once, ever, which is a trial and not a plan. Reading a real meeting means Pro.

The answering half

Under the captions there is a text box.

You type a sentence and press Speak — Ctrl+Enter, or ⌘+Enter on a Mac, does the same thing. Plain Enter inserts a newline, so you can write more than one sentence before it goes. Up to 500 characters at a time, with a counter; over the limit the send button disables rather than silently cutting what you wrote.

If the language you typed in is different from your Speak language, Gaavala translates the message first and then speaks it. So you can type in Turkish and be heard in English, or type in English and be heard in Japanese.

The voice is one of Gaavala's studio voices, or a clone of your own made from a two-minute script you read. If you have used your voice elsewhere — a video, a recording from before — a clone means the room hears something recognisably yours rather than a stock preset.

Send a second message while the first is still playing and it queues behind it. Every message shows its own state in words, not colour alone: Queued, Speaking, Spoken, Cancelled, Partially spoken, Failed. Anything queued or playing can be stopped with one button.

You will not hear your own message. It goes into the meeting, not to your speakers — which for most users is a detail and for you is the whole reason the state labels are written out.

For spoken replies, the Speak Mode guide explains the difference between holding the button and opening a hands-free window.

For a recurring client call, your own voice clone can give Speak Mode a familiar voice; the recording and deletion steps are explained there.

What this needs

This is the section to read before deciding whether Gaavala helps you.

A working microphone connection in the meeting

Typed replies do not ask you to speak or start microphone recording in Gaavala. They still use the meeting’s existing outgoing audio connection, so your device needs a working microphone and the meeting site must have microphone permission.

Teams blocks typed messages while you are muted. Zoom mute also silences the output, but Gaavala cannot read that state. Meet and Webex do not mirror native mute to Speak. Use Gaavala’s Stop control to stop outgoing speech and test with another participant before relying on it.

A Pro plan

Typed messages are part of Speak, which is Pro. The one-time five-minute free allowance covers live captions only and contains no typed speech at all, so it is not a way to try this half of the product. Pro is $19.99 a month for 120 minutes a day.

One of Gaavala's voices

Typed speech runs on Gaavala's studio voices or on a personal clone.

Chrome 138 or newer, for cross-language messages

Translating your typed message happens on your own machine, using Chrome's built-in on-device translator. If that is unavailable, or it does not have the language pair you need, Gaavala refuses the message and tells you — it will never speak your text untranslated and present it as a translation. A message already in your Speak language needs nothing extra.

A meeting in a browser tab

Microsoft Teams, Google Meet, Cisco Webex and Zoom, joined from their web clients in a Chromium desktop browser. Desktop meeting apps are out of reach for a browser extension. On Zoom specifically, the tab has to have been opened with Gaavala already running, or Speak will ask you to reload it.

For the joining and audio steps, use the guide for your meeting: Teams, Zoom, Google Meet or Webex. These explain browser access before you configure typed replies.

What Gaavala cannot tell you

It cannot confirm anyone heard you. Gaavala knows when it has finished sending your message's audio into the meeting. It marks that message "Spoken". That is not the same as delivery, and there is no platform where we can confirm delivery. We would rather you know that than infer a confirmation we cannot give.

The one warning we do have is narrow. On Microsoft Teams, if the meeting microphone gets muted while your message is playing, that message is marked "May not have been delivered". On Google Meet, Webex and Zoom no such warning can appear. The absence of a warning is not evidence that a message landed.

The meeting's mute button does not behave the same everywhere. On Microsoft Teams, muting yourself there silences Gaavala's output with you. On Google Meet and Cisco Webex it does not — those clients churn the microphone track constantly as part of their own audio handling, and mirroring that would chop your speech mid-sentence, so Gaavala deliberately does not mirror it. On Zoom the mute happens inside Zoom's own encoder, where Gaavala cannot read it. On all three, the control that reliably stops Speak is the one in the Gaavala panel.

Typed messages are missing from the record. They are not written into Gaavala's transcript, its TXT/SRT/JSON export, or its on-device AI summary. If you take part in a meeting by typing, your contribution will be absent from the record that colleagues who spoke aloud appear in. That is a gap, not a privacy feature, and we are naming it because someone will otherwise discover it in an action-items list that does not include theirs.

Reading and speaking share one budget. Pro is one allowance of 120 minutes a day, covering both the live captions you read and the messages you type.

The meeting audio privacy guide separates live audio, local transcript history and the optional voice sample.

Why we built the typed half

The honest answer is that it started somewhere else. Speak exists so that someone who does not share the room's language can be heard in it. A text composer was the obvious way to handle the cases where speaking aloud was awkward — a shared office, a noisy room, a sentence you wanted to get exactly right.

What became clear as it was built is that "type it, and be heard in the room" is not a convenience for most people and a necessity for some. For a deaf participant it is the missing half of the meeting: the captions already let you follow it; this lets you answer it in the same channel everyone else is using, at the moment it matters, rather than in a sidebar someone may read later.

If you have tried tools like this before and they failed you in a specific way, we would genuinely like to hear which way. Write to support@gaavala.com.

Billed through Freemius (Merchant of Record). Prices exclude VAT and sales tax — where such tax applies, Freemius adds it at checkout.