Published
New voice model: directions, emotions and lively dialogue
tts
- The Text-to-Speech Studio has moved to a new model with livelier, more expressive delivery.
- Directions in brackets set how a line is played:
[whispering, terrified],[barely holding back tears]. Write them in any language; they are not spoken. - Laughs, sighs and pauses go right into the text where they should sound:
[laughs],[sighs],[pause]. - Paired tags change the delivery of specific words:
<whisper>,<emphasis>,<slow>and more. - Dialogues with up to 4 voices sound connected: lines take each other into account.
- Long texts are voiced about twice as fast.
- Directions and tags no longer count toward the price.
- Old tags still work, and the guide on the service page has been updated.