You can now direct how a voice delivers each line of your script. Type a short cue in square brackets, like [warmly], [slowly] or [with an upbeat tone], and the voice performs the words that follow in that style. Audio tags run on our most expressive voices yet.
How it works
- Pick a voice with the Audio tags label in the voice selector. You can also filter for them under Filters, then Other.
- Type your cue in square brackets right before the words it should shape.
- The tag is highlighted in your script. It shapes the delivery and is never spoken.
- Generate as usual.
You can stack two tags for something more specific, like [thoughtfully][slowly], and use as many as your script needs.
What this means for you
Your script no longer has to sound the same from start to finish. Warm up a welcome, slow down the part people need to remember, and add emphasis where it counts. There's no list of tags to learn. Describe the delivery in plain English, the way you'd brief a presenter.
Good to know
- Audio tags work with voices that carry the Audio tags label. We're adding more over time.
- If a voice doesn't support tags, you'll see a warning and the tag will be read out loud. Switch voices or remove the tag.
- Subtle cues give the most natural result on camera. Avatars don't yet match very intense emotions like shouting or sobbing.
- Results can vary between generations. If a line isn't landing, preview it and try rewording the tag.
Best,
The Colossyan team