Text-to-speech is one of the most requested interaction features on Twitch: viewers type a message, and your stream reads it out loud. A good TTS setup turns a quiet chat into running commentary. A bad one reads spam bots at 2 a.m. This guide covers how TTS on Twitch actually works, the setup options, and the moderation layer most streamers forget.
How TTS on Twitch works
Twitch itself has no built-in TTS. Every TTS setup consists of three parts:
- A trigger. How viewers request a message to be read. Common triggers are Channel Point redemptions, Coins, or chat commands.
- A voice engine. The service that converts text into audio. Quality ranges from robotic system voices to natural AI voices.
- An output. Where the audio plays: a browser source in OBS, or directly on your PC.
The trigger you pick changes the dynamic of your stream. Channel Points make TTS a loyalty reward every viewer can earn by watching. With Coins, messages tend to be rarer and more intentional.
Setting up TTS with ControlPlay
ControlPlay ships TTS as part of the free tier:
- Create your account at controlplay.app and connect your Twitch channel.
- Install the desktop client. It walks you through the full setup, including the OBS output.
- Activate the ControlPlay Twitch Extension on your channel. Viewers can now trigger TTS with Coins directly from the extension panel.
Voices come from a curated voice library. If you qualify for the beta, you can add your own cloned voice. Viewers then speak through a voice that sounds like you, which lands very differently than a generic system voice.
Ghost Voices: TTS through your real microphone
A ControlPlay speciality: TTS doesn't have to play as stream audio. With Ghost Voices, the message comes out of your actual microphone, audible in-game and in voice chats. Your squad hears the message as if you said it. It is one of the most clipped interactions on the platform, because the reaction happens inside the game, not just on the overlay.
Moderating TTS before it goes on air
Unmoderated TTS is a liability: slurs, dox attempts and spam all become audio on your stream. A serious TTS setup needs:
- A profanity filter that blocks before playback, not after.
- Per-viewer timeouts, so one problem viewer doesn't force you to disable TTS for everyone.
- Message length limits, so nobody reads essays into your stream.
All of this ships out of the box with ControlPlay: every message runs through a configurable profanity filter and is blocked before it is ever rendered to audio. Per-viewer timeouts (15 or 60 minutes) handle repeat offenders, and length limits are built in. No extra bots or third-party tools needed.
Which trigger should you use?
| Trigger | Best for |
|---|---|
| Channel Points | Community engagement: earned by watching, high message volume |
| Coins | Priority messages: fewer, more deliberate |
| Both | Most streamers: points for everyday fun, Coins for priority messages |
Start free, see how your community uses it, and tighten the filters as you grow.