Usage guide
How to summon Comeyomi to an X Space, what it reads and what it skips, how to change the reading mode, volume and speed, and how listeners use the web chat.
Last updated
This guide covers summoning the bot, what it reads aloud, and how to adjust it while the Space is running.
The speech is Japanese — see the note on the voice before you start.
Before you summon it
The bot needs the host to approve its speaker request. Posting the summon is not enough on its own, so keep the X interface where you can see incoming requests.
Only the host or a co-host can summon Comeyomi. A listener or speaker posting the mention will not bring it in. If you were named co-host, make sure you accepted the invitation — until you do, you are still a speaker as far as permissions go.
Everything read aloud goes out over the Space audio. Tell your listeners that replies and web chat messages will be spoken, so they can decide what to post. Anything you would not want said out loud — contact details, for instance — should not be posted.
Summoning it
- Reply to the post attached to your Space, mentioning @comeyomi
- Wait for the bot to join. This can take about a minute
- Approve Comeyomi's speaker request
- Send a short message to confirm the audio
For that test message, use something plain and easy to hear. A reply containing only a mention is not useful for testing, because mentions are stripped before reading.
It does not read the backlog. Replies posted before the bot could speak may be skipped, so test with something new after it has become a speaker.
What you hear
Replies and web chat messages
Ordinary comments are read as written. The author's name is not prepended each time, and @usernames inside the text are removed before reading.
Because names are stripped, write comments that stand on their own. "About what you said earlier — a question" works better than a bare mention.
URLs beginning with http or https are replaced with a short spoken marker rather than spelled out letter by letter. The bot does not open links or read the page behind them. When you share something, describe it in the text as well.
Reactions and emoji
X Spaces reactions play a sound effect rather than being read as words. So do some emoji written in the message text.
The two sets are different. The same symbol does not necessarily make the same sound in both places, and symbols with no matching sound are simply ignored.
At most two emoji-triggered sounds play per message.
Joins and speaker requests
When a new listener is detected, the bot announces them by name. Joining and the announcement are not always simultaneous, and repeat announcements for the same person are suppressed.
Speaker requests are announced too. When you hear one, check the request in X. These announcements include names, unlike ordinary comments.
Changing how much is read
Open the Space's web chat and sign in with an X account that has control of the Space. The settings button appears at the top when you have permission and the Space has not ended.
| Setting | What you hear | When to use it |
|---|---|---|
| Full | Comment text plus matching sound effects | Taking questions and reactions |
| SE | Sound effects and a tone, no speech | Talking with other speakers |
| Off | Nothing | Presenting, or when you need quiet |
"SE" is not silence. Choose "Off" for that. Switching to "Off" also stops audio that is already playing.
Comments received while off are not replayed when you switch back.
The same panel has speech volume and reading speed, both on a 1–10 scale, and a separate effect volume from 0% to 200% with a default of 70%.
Turning the speech volume down does not make the reaction sounds quieter — that is the effect volume. In a busy Space, the reaction sounds are usually what feels loud, so lower that first and keep the speech.
Volume and speed apply to the whole Space, so only the host can change them. Anyone can check the current values.
Conversation wait
While someone is speaking, the bot can hold the reading and start once the room falls quiet. This keeps the synthesised voice off the top of your own.
It can be set to on, off, or automatic. Automatic switches based on how busy the Space is.
Comments still queue while it waits. When things are busy, older queued comments may be dropped rather than read late.
The web chat for listeners
Each Space has a web chat at a URL the host receives when summoning the bot. Ask the host for it.
Posting and sending hearts require signing in with X. The chat shows both X replies and web posts together. Links are clickable, and the first link in a message gets a preview card where one can be fetched — so describe what you are linking to in the text as well.
Messages posted from the web chat are not posted to X, and they disappear when the Space ends.
The heart button is a reaction inside the chat. It is not the same thing as an X Spaces reaction, and it does not trigger the effect sounds.
Once a Space has ended, the page says so and no longer accepts posts.
Private accounts
To run a Space from a private account, Comeyomi needs to be following you — X does not let it reach the Space otherwise.
The follow comes with membership. If your account is private, a follow request arrives, and you need to accept it before starting the Space. Public accounts work without any of this.
Reading itself stays free either way.
Running a Space well
Switching by segment tends to work: "Full" while taking questions, "SE" while the speakers talk among themselves, "Off" while presenting.
Say when you switch. If you silently turn it off, listeners cannot tell whether their comments failed to arrive or simply are not being read.
Keep individual comments to one point. They are read in order, but when the queue backs up, older ones may be dropped. Extremely long messages and repeated posts may be skipped as well — so if nothing is being read, check the mode and the bot's state rather than posting the same thing again.
The command list is in the command reference. The Japanese documentation covers considerably more ground: the full guide, the docs section and the FAQ.