Record a voiceover from the chat

Bearly can now turn your conversation into audio you'd actually use. A Voice bar now sits under the latest message in a chat. Select Hear this reply to listen to the answer, or open the bar to record a voiceover, a podcast intro, or a narration: choose a voice by hearing it read your own words, tell it how to sound, and compare three takes side by side.

Start from the job, not the settings

Most people who want speech know what they're making before they know which voice they want. So Voice opens on a short question, What are you making?, with six answers: Video voiceover, Podcast intro, Audiobook, Proofread by ear, Announcement, and Wind-down.

Each one sets a model, a lead voice, two alternates, and a written delivery direction. Video voiceover starts with an informative narrator and asks for a clear, measured pace. Audiobook starts with a warm storyteller and gentle pauses. All of it stays editable.

The script usually comes from the conversation itself. Use reply puts Bearly's latest answer into the script; Use my message puts in the text you sent. Edit from there, or type something new.

Pick a voice by hearing it

There are 30 voices on the Gemini models, and an adjective like "Firm" or "Breezy" only tells you so much. Every voice card has a play button. Select it and that voice reads the first sentence of your script, with the delivery you've written, so you judge it on the words you'll ship. A second tap stops it.

Samples are short and are not saved to the chat. Each new sample is a small generation that counts toward your usage. Replaying a sample you've already heard, with the same words and direction, doesn't make another.

A worked example: a 30-second product demo

Say you need narration for a short screen recording of a new reporting feature. You ask Bearly to draft it, edit the reply, and end up with three sentences:

Every Monday, your team's numbers land in one report. Open it, and the changes since last week are already highlighted. Click any figure to see where it came from.

Open Voice, choose Video voiceover, and select Use reply. The cast reorders so the three suggested narrators come first, each marked with a dot. Tap through their samples, then change the direction to "confident, unhurried, slight smile on the last line."

Now select Compare 3. Bearly records the script in all three voices at once and puts the first finished take in the player. The others land in the list underneath. Play each one. While a take plays, the script highlights word by word, which makes it easy to hear exactly where a pause landed or a word got clipped. Download the keeper and drop it into your video editor.

If none of them is right, change one word of direction and run it again. Every take stays in the chat with the voice and model it used, so you can come back to a version later.

Proofread by ear

Writers and editors have long read drafts aloud to catch what the eye skips. Proofread by ear does that with a neutral voice that reads exactly what's written and pauses at punctuation, on the faster model. Paste the draft, generate, and follow the highlighted words as you listen.

The highlighting is an estimate. Bearly doesn't get word timings back from the speech models, so it spaces the words by their length and punctuation. It follows a steady read closely and can drift on long, dramatic deliveries. Treat it as a reading guide, not a caption track.

Three models

The switch at the top of Voice picks the model:

• Fast uses Google's Gemini 3.8 Flash-Lite TTS. It's the quickest, and suits proofreading and everyday listening.
• Studio uses Gemini 3.8 Flash TTS for more expressive delivery, the better choice for voiceovers and narration.
• OpenAI uses GPT-4o Mini TTS and its own voices.

Gemini takes download as WAV files and OpenAI takes as MP3.

The Read aloud button under each answer keeps its familiar voice. To have it use your Voice settings instead, turn on Use for Read Aloud at the bottom of the bar.

Good to know

Compare 3 records three takes, so it costs three generations. Takes are stored with the chat like other files; download anything you'll need for a long time. Team administrators who have turned off text to speech in a team policy also turn off Voice for those members.

Voice works wherever you use Bearly chat: on the web, in the desktop app, and on iOS and Android. The text-to-speech guide walks through every control.