Stability: Controls how stable the delivery is and how much randomness occurs between generations. Lower values allow a broader emotional range; higher values are more consistent but can sound monotone. Range 0–1. (elevenlabs.io)
Similarity (aka “Clarity + Similarity Enhancement” in the web app): Controls how closely the AI adheres to the original/selected voice characteristics. Higher values push stronger voice-matching; lower values allow more deviation. Range 0–1. (elevenlabs.io)
Style: “Style exaggeration” that amplifies the speaker’s style. Non‑zero values can increase latency, and for some voices may introduce instability (e.g., inconsistent speed or mispronunciations). Range 0–1. (elevenlabs.io)
These sliders are available per track in Dubbing Studio’s track voice settings (you can set them at track or clip level). (elevenlabs.io)
Stability: Controls how consistent vs. variable the delivery is from one generation to the next. Lower stability increases randomness/emotional range; higher stability is more consistent but can sound flatter/monotone. It maps to the “Stability” slider. (elevenlabs.io)
Style: The “style” slider is style exaggeration. It amplifies the speaker’s characteristic style/performative traits. Raising it can increase latency and may make the model slightly less stable; recommended default is 0 unless you specifically want stronger stylization. (elevenlabs.io)
In Studio/Dubbing Studio, these appear in Track Voice settings; you can also override settings per selection if needed. (elevenlabs.io)
shy wren
main girder