Text-to-Speech

Automate → Text-to-Speech controls who can speak, what gets filtered, and how chat messages enter the voice queue. The pill in the top-right corner sums up the current setting, for example Reading for followers.

Notices

The top of the page lists anything you should know about the current setup, such as:

• Messages leave this computer to be spoken by a network service — shown when you use a network voice such as ElevenLabs.
• Platform limitations — see Platform notes below.

Who may be read aloud

Pick one of Nobody, Followers, Subscribers, Moderators or Everyone.

Nobody is the off switch. Everything else is a floor, not a list: choosing Subscribers reads subscribers and everyone above them, such as moderators. Everyone reads all of chat.

Speech playback

Speech plays through Windows’ default audio output. The pitch and rate depend on the installed voice.

• Test voice plays a sample.
• Skip current skips the message being spoken.
• Stop all stops all speech.

Voice engine

Choose where the speech is made, and what that costs you:

Engine	Where speech is made
System	The voices installed in Windows.
On-device	On your PC.
ElevenLabs	ElevenLabs, a network service billed per character. Needs the network and an API key.

ElevenLabs accounts

You can save one or more ElevenLabs accounts and use their voices:

1. Enter an account name (for example Main or Character voices).
2. Paste the account’s API key. It’s stored in Windows Credential Manager.
3. Press Save account.
4. Press Load voices to fetch that account’s voices.

Use Update key to replace a key, or Remove to delete the account from ilyStream. Spoken text is sent to ElevenLabs and uses that account’s credits.

Platform notes

• YouTube doesn’t tell apps who subscribes to a channel, so when you choose Followers or Subscribers, only YouTube members and moderators are read.
• Kick viewers are read once they follow during the stream.

Related

• AI Co-Host can speak with the same voices.
• ElevenLabs integration: what data it uses.