Module livekit.agents.voice.presets

Expressive presets (framework-internal, not publicly exposed).

A preset is a use-case (customer service, casual) that is provider-agnostic: each presets.* constant is just an ExpressiveOptions carrying a preset. When expressive mode is active the framework resolves it against the active TTS provider (via tts.markup._provider_key()) and injects the variant tuned for that provider's markup tags. A provider with no tuned preset falls back to the agnostic default, which still injects that provider's tag reference through the {tts.markup.llm_instructions} placeholder — so a preset always does something sensible and can never disagree with the markup pipeline (both read the same provider key).

Classes

class Preset (*args, **kwds)

Expand source code

class Preset(enum.Enum):
    """The domain a preset is tuned for. Used to key the per-provider registry."""

    CUSTOMER_SERVICE = "customer_service"
    CASUAL = "casual"

The domain a preset is tuned for. Used to key the per-provider registry.

Ancestors

• enum.Enum

Class variables

var CASUAL

var CUSTOMER_SERVICE