Here’s a quick tour of the settings panel in ElevenLabs Text to Speech, and what each one does.
The settings panel
The settings sit on the right of the Text to Speech page. From the top: the voice, the model, three sliders for stability, similarity and clone strength, then language, audio effects and the output format.
Stability
Hover over Stability and ElevenLabs explains it: it sets how adventurous the model can be. Toward Creative you get expressive, varied phrasing and tone. Natural is a conversational balance. Toward Robust it’s precise, stable and predictable. The default sits in the middle, at fifty percent.
Similarity and clone strength
Similarity runs from low to high and defaults to seventy-five percent. Clone strength sits below it, also from low to high. If a read sounds off, change one slider at a time and listen again, so you know which change made the difference.
Language and output format
Language is set to Auto by default. Output format is where you choose the audio file you get back. Here it’s MP3 at forty-four point one kilohertz, one hundred and twenty-eight kilobits per second.
The model
And the model picker, near the top, changes which ElevenLabs model reads your text. I cover choosing between them in a separate guide. Whatever you pick, the reset values button puts the sliders back to their defaults.
Hear the voices yourself
Every voice on AI Voice Picks reads the same script, so you can compare them directly. Ready to try it? Start with ElevenLabs (affiliate link).