Edgewise API
    Preparing search index...

    Interface SpeakOptions

    Options shared by every verb.

    interface SpeakOptions {
        allowPreview?: boolean;
        device?: DeviceOption;
        dtype?: Dtype | Record<string, Dtype>;
        exaggeration?: number;
        input: string | AsyncIterable<string, any, any>;
        model: ModelRef;
        onProgress?: (event: LoadEvent) => void;
        signal?: AbortSignal;
        speed?: number;
        voice?: VoiceSpec;
    }

    Hierarchy (View Summary)

    Index
    allowPreview? device? dtype? exaggeration? input model onProgress? signal? speed? voice?
    allowPreview?: boolean

    Allow preview and experimental models.

    device?: DeviceOption

    Force a device. Default 'auto'.

    dtype?: Dtype | Record<string, Dtype>

    Force a dtype, such as 'q4'.

    exaggeration?: number

    Cloning models: how expressive the speech is, 0 to 2. Default 0.5.

    input: string | AsyncIterable<string, any, any>

    Text, or a stream of text deltas such as a running generate().

    model: ModelRef
    onProgress?: (event: LoadEvent) => void

    Receives load events while the model downloads and compiles.

    signal?: AbortSignal

    Abort download or inference.

    speed?: number

    0.5 to 2. Default 1. Kokoro only.

    voice?: VoiceSpec

    Settings
    Member Visibility

    On This Page
    Properties