Bright Oddity Press
Audiobook Studio 1.x docs

The Voice Lab

The Voice Lab is where you build your narrators: a voice for each reader or character, cloned from short recordings.

At a glance
  • A Voice is an identity (for example "Dracula").
  • A Variant is a style of that voice (for example "Angry" or "Calm"), built from its own recordings.
  • Add 3 to 5 clean .wav samples, then build.
  • Open it from the microphone icon in the top bar.

Voices and variants

  • Voice: a narrator or character identity. Every voice has at least one variant.
  • Variant: a style or mood of the same voice, such as Normal, Angry or Whisper, each built from its own recordings.
  • Samples: the recordings used to clone the voice.

Create a voice

  1. Click + New Voice at the top and give it a name.
  2. Open the voice's card. Opening one card closes the others.
  3. Drop 3 to 5 clean .wav files into the Samples area. (You can also use the add button.)
  4. Click Rebuild if the voice needs it, then Generate Sample to hear a preview.
  5. To add a style, click + Variant and give it its own samples.

The Voices page with a Dracula voice open, showing Angry and Calm variant tabs, a play button, speed, Script and Rebuild buttons, and three sample files

Tune and test

  • Speed: set the default speaking rate between 0.5x and 2.0x.
  • Script: change the text used for the preview clip.
  • Samples: the first sample shapes the voice most; later ones add nuance. Mixing clean clips with different delivery styles can give a richer voice.
  • Update indicator: a small turning arrow on a voice's avatar means a variant needs samples or a rebuild.
  • Voice menu: the three dots on a card let you set the default voice, rename a voice, or delete it.
Heads up: Deleting a voice removes all of its variants and sample files from your disk.

Each voice can use its own engine. XTTS (Local) is the default; the optional cloud engine is explained in Settings.

Sharing a voice

Each voice keeps its own files in its own folder (a profile, a built voice file, and a preview), so renaming, moving or sharing a voice is safe. A lightweight starter voice only needs the profile, the built file and an optional preview, not every original recording.

Next