Keeping text-to-speech spam-free on TikTok Live
Practical moderation settings for streams that read chat aloud — cooldowns, rate limits, filters and what to do during a raid.
Text-to-speech changes the stakes of moderation. A message that would scroll past unnoticed in a text chat gets read aloud to everyone watching. Anything you would not say on stream yourself needs to be caught before it is spoken.
The three limits that do most of the work
- Per-user cooldown: a minimum gap between messages from the same viewer. This alone stops the most common form of TTS abuse, which is one person spamming to hold the audio channel.
- Global rate limit: a ceiling on messages per minute across everyone. This is what protects you during a sudden influx.
- Message length cap: long messages are both tedious and the usual vehicle for abuse.
Filters
Maintain a blocklist of words that are never spoken. Include the obvious slurs, but also add terms specific to your community that get weaponised, and any phrasing that sounds like an announcement from you rather than from a viewer — impersonation is easy when everything is read in the same voice.
Per-viewer voices help here too. When a specific username always sounds the same, an impersonation attempt is immediately audible as coming from the wrong voice.
Surviving a raid
- Have a way to mute TTS instantly without ending the stream. Know where that control is before you need it.
- Tighten the global rate limit rather than turning speech off entirely — you keep the feature while cutting the volume.
- Restrict speech to followers or subscribers temporarily if the influx is hostile.
Give your moderators a role
Assigning a distinct default voice to moderators makes their messages instantly recognisable, which matters when they are trying to reach you during a busy moment. It also means viewers can hear that a correction came from a moderator rather than from a random account.