Inside the studio
With screenshots. The home page is the short version.
Casting
The cast list shows every speaker found in your manuscript. Each character gets a color, a voice and a portrait from a gallery of 445. Pick a portrait in Character Tags, or press Best fit auto assign for the whole cast at once. A character with no voice yet is flagged, and so is a line whose speaker is in doubt.
Each card shows the character’s voice and portrait and how many lines they have. The microphone button on the card opens their part for recording.
Recording
It scrolls at the pace you set, the current line held in a highlighted band, and your hands stay off the controls. Stop, reread a sentence and carry on. A session runs from a page to the whole book. When you stop, the reading is divided into the manuscript’s lines by the words you read, on your own computer, and nothing is sent anywhere. Each piece is filed under its line with its waveform and length.
Each take is also compared with its line. A word you left out or changed is marked on the take, and you say whether that was on purpose or a mistake. A single line is recorded the same way, so you read the chapter straight through and then re-record the lines that need it.
StellAurator opens your microphone through the driver that came with it, and you hear yourself in your headphones with no processing added. Your other programs keep the headphones too; nothing else is muted while StellAurator is open. A Reduced delay switch takes the delay down to the shortest your hardware supports. Which microphone to buy, by budget, is in the microphone guide; the interfaces and the headphones have their own guides.
Drop in a chapter recorded in another program and choose the chapter, and the file is divided into your lines by the words, on your own computer, or at its pauses. The file itself is never changed. Where the recording and the manuscript disagree, you decide, line by line.
Cast as many actors as the book needs. Each actor gets an invitation link and a free copy of StellAurator that shows only their lines, scene by scene, with your direction above the script. The line before each one plays as a cue, and their recordings come back onto their lines, where you accept each one or mark it for another take. The production stays on your computer.
Direction
Direction is the note a director gives an actor before a take. It is written once on the line it belongs to, and whoever records that line sees it: you, an invited actor or a voice service that takes direction. It is kept apart from the manuscript text and never changes a recording already made.
Direction is written for a scene, a character within it, a beat where the scene turns, or a single line. It stays on screen above the script while the take is recorded. An actor recording at home reads it there, and so do you when you come back to a chapter.
Drag the character’s feelings into place on the emotional web, with up to nine feelings at once, and set the delivery on eight faders, from urgency to warmth. The blend is named (Bittersweet, Grim resolve) and a direction note is written from it. Twelve presets set a mood with one click, and narrator lines have quick chips. A note you wrote yourself is never overwritten. One such note: “Attend closely while fear and strain sharpen awareness.”
With a text AI service connected, every line gets a mood from the scene so far and from who the character is at that point in the book. It reads only what comes before the line. Every line is a separate undo step, and lines you directed by hand are left as they are. Direction also goes to the voices that take it. ElevenLabs v3 and v4 voices get the mood as a short tag in front of the line.
Clean-up
Each recording is checked for clicks and pops, for breaths, for hum and for background noise, and a fix for each one is listed under it. Review plays the recording before and after, with every fix marked on the waveform. Drag across a breath you want to keep and that fix is skipped. Apply keeps the set.
A fix is a layer over the recording, turned off or removed later, and the original is never changed. An icon on each recording shows what has been applied; click it for the list and for where each fix sits.
Each line keeps every recording made of it, with its waveform and length, until you delete one, and you pick the one to use. A spool is one complete reading of the book, made from the recording picked for every line.
Set once for the book, these apply to every take as it lands: a rumble cut for low-frequency noise, a softener for harsh S sounds and three tone controls. One gain setting covers every recording made on this computer, and each chapter’s loudness is set when it is exported.
Spools A and B are two columns beside the lines. Pick a different recording in each where you want to compare, play either through, and export either. You add more spools as you need them.
Fixing by hand
The suggestions are made from these tools, which also run on one recording or on every selected line. They are grouped on the right; choosing one opens it under the recording, with its settings, before-and-after playback and Keep.
The timeline and the music
With an AI service connected, Auto Score reads each chapter and marks where music belongs, such as where a scene turns or where the tension peaks. Each cue gets a mood and a one-sentence reason, and it is tied to the words it belongs to, which keeps it in place when the chapter is edited.
You give the book a musical identity first, in your own words or by naming films and games whose music is close to what you want. Your own music and sound effects go under the lines from the sound library, and a service you connect generates music and effects to match that identity. Music runs under a scene, and a short hit or a sound effect sits between two lines. The list of music cues prints out for a composer.
Finishing and saving
You hear the chapter as it will be exported, pauses included, and each line is ticked as it is heard. Press F on any line to mark it for re-recording without stopping playback. When every line has been heard, you sign the chapter off. If a line is changed later, only that line needs hearing again.
Choose the output format: Clean narration, Audiobook master for Audible’s ACX, Podcast, or your own recipe. The Finish screen then runs the chapter, or the whole book, through five steps. Diagnose measures every recording and reports what it found, in words and numbers. Plan lists what will be done. Apply does it, as removable layers. Check measures the result against the requirements. Export writes the files.
One MP3 per chapter in ACX’s format, one M4B with chapter marks for other stores, or both. You choose which recordings go in, the picked ones or a whole spool. The pauses between lines are measured on export and set to the lengths you chose for a sentence end and a paragraph break. A muted track is left out of the export, and the export window says so first.
ACX does not accept computer-generated narration. A reading you recorded yourself and then converted into a character’s voice is still your performance, and the original recording is kept beside every converted one. A file with a computer-generated voice is marked as containing AI-generated narration, and a file with a reading you recorded and converted is marked as containing a speech-to-speech modified human performance. The export window counts the generated and converted lines before you export.
Your computers
Sign in on a second computer and the first one asks whether to let it receive your books. After that the two computers are paired, with no code to type. You send a book straight to another of your computers, or keep it in sync across them. While your main computer is on, you open its library from the other one to listen and read your notes.
Technical details
64-bit Windows 10 or 11 today; a Mac version and phone apps are on the way (get notified). The installer is signed, so Windows shows the publisher when you run it. Everything except the optional services works without an internet connection.
Manuscripts as Word, PDF, EPUB or text files. Recordings as mp3 or wav files, among other common formats. Finished audiobooks as one file per chapter in ACX’s format, or one M4B file with chapter marks. A whole book, with its cast and its recordings, as one backup file you open on another of your computers.
Twelve themes restyle every screen, all included. See the themes.
No card. A practice book and a walkthrough are included. Exporting the finished audiobook comes with a plan.