Outputs and optional voice conversion

Choose the output that matches the project. An audiobook joins narration in reading order. A synchronized voiceover follows a recording's timeline. A narrated video book presents book text with generated narration; it is a different task from dubbing an existing film.

Audiobook output

In Output, choose the selected generation/audio version, format, bitrate and book metadata. Set the cover where needed. M4B is useful for book-oriented players; WAV or FLAC can be useful for further editing. Test the actual player and device you intend to use rather than assuming every format preserves every feature.

Assemble a short section first. Check chapter boundaries, pauses, cover, title and the first and last spoken words. If you change metadata or output encoding, export again; you do not need to synthesize unchanged speech.

Video book and voiceover

For a video book, review how the written text is presented alongside the narration and preview a short export. For a voiceover, first inspect source timing, the chosen written caption revision, original soundtrack level and generated speech mix. A correct audiobook take is not automatically synchronized to a film.

Choose selectable or burned captions according to where the file will play. Burned captions are part of the picture; selectable tracks need player support. Open the completed export outside Pandrator and seek near the end as well as the beginning.

RVC after speech

RVC is an optional conversion stage that changes the voice of existing synthesized speech. It requires its own installed/running service and compatible voice model. It is separate from choosing a TTS voice or cloning a voice from a reference.

Keep a dry speech export and compare a short converted selection before processing more. Listen for consonants, noise, pitch changes and timing. A conversion cannot be assumed to repair an omitted or mispronounced word: correct that in speech generation first.

Optional XTTS training is a separate, model-specific preparation task. It is not required to use a saved reference in a cloning-capable model. Check the installed component's prerequisites, voice-recording permissions and training controls before starting; model availability does not prove a training service is ready.

See supported formats, local audio tools and generation and takes.