How to make an AI voice sound emotional
Short answer
To make an AI voice sound emotional, use a text to speech model that supports emotion tags and write the tag in square brackets before the line, for example [whispers] Come closer. In VoiceFlow AI, choose the v3 model, which is open to everyone, add a tag such as [excited], [sad], [laughs] or [sarcastic], and generate. The voice performs the emotion instead of reading the tag out.
Step by step
- Install VoiceFlow AI and open Text to Speech.
- Pick the v3 model and a voice. Tap voices to preview them.
- Write your line and put a tag in front of it: [whispers] I have a secret.
- Tap a tag chip in the app to insert it, or type your own.
- Generate, listen, and move or swap tags until the delivery feels right.
Tags that work well
Feelings
[excited][sad][curious][sarcastic][mischievously][cheerfully][crying]
Delivery
[whispers][shouts][laughs][laughs harder][giggling][sighs][exhales][stuttering][wheezing][snorts][swallows][gulps][woo]
Sounds around the voice
[applause][clapping][gunshot][explosion][leaves rustling]
Experimental
[sings][strong French accent]
Tips for natural delivery
Use one tag per line, at the start. Let punctuation carry the rest: an exclamation mark lifts the energy, an ellipsis slows it down. Save the strongest tags such as [shouts] or [crying] for the moments that need them.
Hear it
One voice, any emotion
- [excited] We did it! The whole thing is finally ready, I can’t believe it!
- [whispers] Come closer. I’ll tell you a secret, but you can’t tell anyone.
- [laughs] No way, you actually wore that to the meeting? That’s brilliant!
- [sarcastic] Oh, great. Another Monday meeting. Exactly what I was hoping for.
- [sad] I kept the light on all night, but you never came home.
- [shouts] Hey, wait! Don’t you dare leave without me, I’m coming!
Untouched clips from the app. Voice: Hope, v3 model.
21 more tags in the app
Several voices, one take
Same line, three voice models
“Wait, you made this yourself? Honestly, it sounds just like a real person.” Flash is fastest, v2 the most lifelike for long reads, v3 the most expressive. All three are open to everyone.