pipertts - v1.3.0
    Preparing search index...

    Interface NativeSession

    High-level inference session: phoneme ids + scales to float32 audio (voice.py:498-575).

    interface NativeSession {
        close?(): Promise<void>;
        run(
            feeds: {
                phonemeIds: number[];
                scales: [number, number, number];
                speakerId: number | null;
            },
        ): Promise<Float32Array<ArrayBufferLike>>;
    }
    Index
    close? run

    • Releases native session resources, when supported.

      Returns Promise<void>

    • Runs one inference and returns mono float32 audio samples.

      Parameters

      • feeds: {
            phonemeIds: number[];
            scales: [number, number, number];
            speakerId: number | null;
        }
        • phonemeIds: number[]

          Phoneme ids including BOS/EOS padding.

        • scales: [number, number, number]

          [noiseScale, lengthScale, noiseWScale].

        • speakerId: number | null

          Speaker index, or null for single-voice models.

      Returns Promise<Float32Array<ArrayBufferLike>>

      Raw audio samples at the voice sample rate.

    Settings
    Member Visibility

    On This Page
    Methods