Voice studio · on this machine
Every avatar, any mood.
Type a script, tap a voice and it speaks. Change the mood on the monitor, push it harder or softer, or drop [tags] into the script to change mood mid-sentence. Write Emma: and George: lines for a conversation.
1 · Who speaks
Tap a voice to choose it and hear your script straight away.
2 · The script
3 · Speak it
First sound—
Length—
Sentence—
Voice—
Engine—
Design a new voice
Blend up to four voices into one of your own, then name it and keep it. It shows up under Designed and can be used in any order or editor scene.
What this does, honestly
- Instant start. The script is split into sentences. The first one is spoken as soon as it's ready, usually in a few hundred milliseconds, while the rest are made in the background. The First sound figure is measured on each play.
- Two kinds of engine. Instant is the house voice with the mood engine shaped round it. Emotional · Chatterbox and Emotional · Zonos are voice models that take the emotion in and act the line, using each presenter's own voice from a reference recording. They take about a second per sentence once warm, and the first line after a quiet spell takes longer while the model loads onto the graphics card.
- Moods on the Instant engine change the delivery: speed, pitch, how much the pitch moves (the biggest difference between bored and excited), the brightness of the voice, breath, whisper, tremble, pauses and loudness. The voice model itself has no emotion input, so this is a strong delivery change rather than an actor feeling the line.
- Tags such as
[excited],[whispers],[sad],[laughs],[sighs]and[pause]work the same way as ElevenLabs' audio tags. A mood tag lasts until the next one. - Repeat replays the last take straight from memory, with no waiting.
- Everything runs on this machine, free. Cloned voices and dubbing into other languages stay on the paid route, which needs approval.