Skip to content

Audio Settings

The Audio tab controls how RecapSynth transforms your generated narration scripts into spoken voiceover tracks. You can use local neural models that run completely offline without API fees, or connect high-fidelity cloud voices.

Audio configuration panel showing Enable Audio toggle, engine model selection, voice actor picker, and speech rate slider
The Audio tab: toggle voice generation, choose local or cloud TTS models, and tune speech cadence.

  • Enabled (On): Generates voiceover audio for each completed chapter script, inserts natural breath pauses between sentences, and creates time-synchronized .srt subtitle files.
  • Disabled (Off): Skips the audio generation phase entirely. The processor only outputs scripts and cropped panels, saving credits and processing time if you plan to record your own voiceover or upload external audio.

RecapSynth integrates both zero-cost local models and cloud providers:

Aether TTS (Kokoro · Local)

Runs locally on your machine via ONNX Runtime. High-quality natural speech with zero external API calls or credit deductions. Downloads model weights automatically on first run.

VoxCore TTS (Supertonic-3 · Local)

High-speed local neural synthesis operating at 44.1kHz. Uses 10 distinct male and female preset voices optimized for narration, audiobooks, and commentary.

Edge TTS (Cloud)

Cloud-based natural neural voices with wide language and accent coverage. Requires an active internet connection but no private API key.

ElevenLabs (Cloud)

Studio-grade generative voice AI supporting realistic emotion, pacing, and cloned character voices. Requires an ElevenLabs API key saved in Settings.


Once an engine is selected, the Voice Actor dropdown lists available voices with gender indicators (Mars/Venus icons), language tags (e.g., en-US, en-GB), and scenario recommendations (e.g., Audiobook Narration, Dramatic Intro).

Adjust the Speech Rate slider to control narration speed:

  • 0.1x – 0.8x: Slower pace suitable for solemn story reveals or emotional flashbacks.
  • 1.0x (Default): Natural conversational pacing.
  • 1.2x – 3.0x: Accelerated pacing ideal for fast-paced action recaps and short-form video formats.

Whenever voiceover is generated, RecapSynth automatically runs an embedded speech-recognition model on the output audio to produce a word-aligned .srt subtitle file.

These subtitle files are linked to the chapter’s voice track, allowing you to preview subtitles directly in the Review player and export synchronized text layers in video editors.


Toggle Override Previous Audio if you have adjusted speech rate, switched voice actors, or edited dialogue and need to regenerate audio for chapters that were already processed. When turned off, chapters with existing .wav files are skipped to save processing time.