Choose your AI voice. Each personality runs its own ambient audio layer and accent color.
Pick a preset or author your own steps. Each step has a trigger phrase, an action label, and a spoken response you will hear in the active personality's voice.
Voice Director is a browser workflow rehearsal tool. Each step stores a trigger phrase, an action label, and a spoken response. Running the workflow reads responses in order. It does not listen for triggers, open pull requests, or execute the action labels. The code preset has three steps. The runner highlights each step and awaits its speech completion. Before the second and third responses it waits three hundred fifty milliseconds. That adds seven hundred milliseconds between responses, apart from synthesis and transition time. The default mixer uses rate zero point nine five and pitch zero point eight. Rate can range from zero point six to one point four. Pitch ranges from zero point five to two. These values are assigned to the browser speech utterance. The diagram shows the settings, not a measured waveform or guaranteed duration. Six named personalities change presets and presentation. The source chooses the first installed voice matching the personality language, or the first available voice. It does not clone a fictional actor. Export writes the mixer and steps to JSON and Markdown. Keep these scripts separate from a real agent execution system.
This will clear the saved workflow, mixer values, and the active personality, and return the studio to the default JARVIS state. This cannot be undone.