Voice Notes in Chat
Sometimes you don't want to read the answer. You're walking to a meeting, your hands are busy, or the explanation is simply easier to take in by ear.
Ask AI Chat for a voice note and it replies with a short spoken message, right there in the conversation.
Asking for One
Say so in the message you want answered aloud:
- "Send me a voice note on this."
- "Say that out loud."
- "Can I listen to this instead?"
- "Explain the second section as a voice note."
That's the whole trigger. Voice notes are only ever sent when you ask for one in that message — chat never decides on its own to answer with audio, and asking once does not turn every later reply into a voice note. Ask again whenever you want another.
Playing It
A voice note arrives as a small player in the reply:
- Play / Pause to listen.
- Scrub to any point to skip or re-hear a part.
- Show transcript to read along, and Hide transcript to put it away again.
The text is there deliberately: a spoken answer you cannot search or skim is worse than a written one, so every voice note carries its words too.
Because the reply is spoken, chat writes it to be heard — conversational sentences, symbols and units said the way a person would say them, no headings or bullet lists. It won't repeat the whole note underneath as text either; you usually get a one-line lead-in and the player.
Voice and Language
A voice note uses your assistant voice — the same one read-aloud and Voice Mode use. Change it from the voice bar in Voice Mode or in Settings → Chat, and voice notes follow.
It also speaks your language. If you are chatting in Japanese, Spanish, or Arabic, the note comes back in that language rather than in English with an accent. See Languages in Scholarly.
Length
A voice note is a short spoken reply — up to roughly three minutes, and one per answer. That is deliberate: it is the audio equivalent of a chat message, not an episode.
When you want something longer and more produced — two hosts, chapters, something to listen to on a commute — make an AI Podcast instead. Ask chat for "a podcast on this" and it starts one as a background creation.
| You want | Ask for |
|---|---|
| This answer, out loud, now | A voice note |
| A document read start to finish | Text to Speech |
| A produced, multi-speaker episode | An AI Podcast |
| A back-and-forth spoken conversation | Voice Mode |
Keeping a Voice Note
Voice notes also appear in the conversation's Files — the Files button in the full-screen chat header — so you can download one as an audio file and keep it, send it on, or listen to it outside Scholarly. See AI Chat.
What It Costs
A voice note is part of a chat reply, so it does not use an AI Creation Credit — the same as every other answer chat gives you. Your normal chat message allowance applies, exactly as it would for a written reply. See Plans and Limits.
If It Doesn't Arrive
Occasionally you'll get a written answer when you asked for audio. The usual reasons:
- The request wasn't in that message. Asking in an earlier message does not carry forward — put it in the message you want spoken.
- The answer was too long to speak. Ask for a shorter version ("just the three takeaways, as a voice note").
- The audio couldn't be produced. Chat answers in text rather than leaving you with an empty player. Ask again, or shorten the request.
If a player appears but shows Couldn't load audio, reload the conversation — the note is stored with the message and comes back with it.
Related
- AI Chat — the full chat guide.
- Voice Mode — a hands-free spoken conversation, rather than one spoken reply.
- Text to Speech — a whole document read aloud and saved as audio.
- AI Podcasts — produced, multi-speaker episodes.
- Languages in Scholarly — interface, AI output, and spoken language.
- Chat History and Threads — finding the conversation again.