Text to speech (TTS) reads out a sentence you write, filled in with live details: the viewer’s name, the gift name, the text of their message. Two killer uses: thank every gift without taking your eyes off your game, and let your whole community hear the chat.
Two voice engines
- Edge TTS v2: Microsoft Edge’s neural voices, far more natural, in many languages. It needs an Internet connection. It is the default engine on a new action.
- Legacy TTS: the voices installed in Windows. It works offline, but sounds more robotic.
Add a voice to an event
- 1Open the Events page and create an event, for example Gift received · Rose.
- 2Add the Text to Speech reaction, in the AUDIO category.
- 3In Message, write the sentence to read. Under Available Variables, click a variable to copy it, then paste it into your sentence.
- 4In the Edge TTS v2 tab, type en-US in the Voice field to show only US English voices, then pick one.
- 5Set Volume, Speed and Pitch, then click Test.
Variables are replaced when the sentence is read: {nickname} (the display name, the most natural one to hear), {username} (the TikTok handle), {message} (the comment text, for a New message event), {giftName} and {giftCost} (for a Gift received event). A variable that does not exist for that trigger is not replaced: it is read as is.
Voice settings
- Volume: from 0 to 100, to sit the voice under your music or game.
- Speed: from -100% to +200% with Edge TTS v2, from -10 to 10 with Legacy TTS.
- Pitch: from -50 to +50 Hz, with Edge TTS v2 only. Lower for a deep voice, higher for a cartoon voice.
- Random voice: tick it and pick a Language. Each reading uses a random voice in that language, with a random speed and pitch (±10% / ±28 Hz).
Recipe 1: thank every gift
- 1Create a Gift received event and pick the gift, or use * as the ID to react to every gift.
- 2Add Text to Speech with a thank-you sentence that uses {nickname} and {giftName}.
Watch out for combos: every gift in a combo triggers the event. Thirty roses in a row means up to thirty thank-yous. With *, keep the sentence short, or save the voice for rarer gifts.
Recipe 2: a different voice for each gift
Create one event per gift, each with its own Text to Speech action: a soft voice for the Rose, a deep voice with Pitch all the way down for the Lion. If you also keep a * event for all gifts, a Rose triggers both: remove the voice from the * event so you do not hear it twice.
Recipe 3: read the chat
- 1Create a New message event and put * in Chat command: every chat message triggers the event.
- 2Add Text to Speech with {nickname} and {message} in the sentence.
- 3Set a Cooldown (s), and turn on Per-viewer cooldown so one viewer cannot hog the voice.
- 4To stay in control, turn on Followers only or Subscribers only.
You can also answer a specific command, like !hello. The command must be the whole message: !hello triggers the event, !hello how are you does not. To read free text, use * and filter who can trigger it.
Sentences to copy
Thanks {nickname} for the {giftName}!{nickname} says: {message}Hi {nickname}, welcome to the live!How the voice leaves your PC
- NexusTok plays the voice on the Windows default audio output: your headset or your speakers.
- Sentences play one after another, without cutting each other off. The voice has its own queue: a sound and a sentence can therefore overlap.
- When many sentences arrive at once, some may be skipped instead of piling up.
- Emojis are removed before reading. Links and the rest of the text are read as is.
- The Stop sounds button, on the Dashboard, clears the queue of pending sounds and sentences. The sentence being read finishes.
If it does not work
- You hear the voice, your viewers do not: your streaming software is not capturing your PC sound. In OBS, check that the Desktop Audio source is on, not muted, and set to the same device as the Windows default output. You can also add an application audio capture that targets NexusTok.
- Nobody hears anything: with Edge TTS v2, check your Internet connection, or try a voice from the Legacy TTS tab. Also check the action’s Volume.
- The voice talks too much: raise the chat event’s Cooldown (s) and turn on Per-viewer cooldown. The cooldown only applies to chat events: for gifts, target specific gifts rather than *.
- Things go off the rails mid-live: click Stop sounds to clear the queue.
Is text to speech free?
Yes. The action is not Pro-only: it simply counts toward the free plan’s event limit.
Can it read messages in another language?
Yes: pick a voice in that language, or Random voice with the Language you want. A French voice reading English keeps its accent.
Do I need to install voice software?
No. Edge TTS v2 works over the Internet, Legacy TTS uses the voices already in Windows.
Set it up on your next live
Install NexusTok, connect your live, add your first overlay to OBS.
Download for freeFree to start, no credit card, Windows 10 and 11