Olee Voice Studio

OpenAudio S1 · Tesla T4

connecting…
GPU
VRAM — / —
RAM — / —
GPU —% CPU —% TEMP —° PWR —W
01 Compose
0 / 200000
Expressions tap to insert · use ( ) not [ ]
a fixed voice stays consistent across every generation
Style
Faithful = most accurate clone · Balanced = default · Expressive = most natural/emotional (less faithful)
Pacing
Natural/Relaxed = pause between paragraphs · Smart = an AI places natural pauses · Script = paste an ElevenLabs-style script ([WHISPER], [PAUSE 3s]) — cleaned, generated as continuous passages, cues performed
Pause scale 1.0× Script mode — scales the director's natural per-phrase pauses ([PAUSE] cues stay exact). 1× = as directed · higher = slower, more meditative.
Speed 0.90× Speaking rate (pitch unchanged). Lower = slower, calmer — closer to a natural narrated read.
Format
Temperature 0.80 stability ←→ expression
Top-p 0.80 focus ←→ variety
Seed
fix it to reproduce a voice
Voice cloning off

Drop a reference clip or browse

10–30s of clean speech works best
02 Output

No audio yet

Your generated voice will appear here

History