A multicast audiobook is usually not recorded with the whole cast in one room. Each actor records their own lines, often in their own booth on their own schedule, and someone then builds every scene from those separate takes. Publishers Weekly described the format years ago as created by editing together individual voice performances, and that assembly is still a long editing job: finding each line in each file, trimming it, and placing it after the line before.
Cast assembly does that part. You upload each actor's recording for a chapter, and every line is cut from their take and placed where it belongs in the book.
It is one of Midsummerr's Pro services, built for production houses, audio-drama producers and authors who recorded their book with a cast.
What it does
You bring two things: the manuscript and the recordings. The manuscript tells us who speaks every line; each actor's recording supplies the performance.
For every chapter, you attach one recording per character. The narrator is a character too, so their file goes in the same way. Each line is cut out of the right actor's take and placed in the order the book tells it, so the chapter plays as one scene rather than a pile of separate sessions.
The actors' takes are the audio. A line that came from a recording is never re-voiced, never directed and never smoothed over by a generated voice. If the actor read it, that reading is what the listener hears.
Ready to try it on your own book?
Start with 5,000words free →When a character has no recording
Real casts have gaps. The actor playing the innkeeper hasn't sent their file yet, or nobody was ever booked for the three lines a messenger has in Chapter 9.
Cast assembly never fills those gaps on its own. A speaking character with no recording shows as No recording in that chapter's Recordings list, and the chapter waits. You have two ways forward:
- Upload the recording when it arrives.
- Approve TTS for that character in that chapter, and an AI voice speaks their lines there.
Approval is per character and per chapter, so you can give a walk-on part an AI voice in one chapter and still wait for the actor in the next. Generate all skips any chapter that is still waiting and tells you why, and nothing is charged for a chapter that hasn't generated.
How an order runs
The order lives in the same book layout every Midsummerr production uses.
| Step | What you do | What you get |
|---|---|---|
| Order | Upload the manuscript, title and language from Pro services | Every line assigned to the character who speaks it |
| Voices | Review the cast list | Every speaking character, with how many lines each one has |
| Recordings | Attach each actor's recording to their row in the chapter | A tick per character as the files come in |
| Generate | Generate the chapter | The assembled chapter, ready to review in the studio |
Recordings can be WAV, MP3, M4A, FLAC, AIFF, OGG or WebM. Because the order starts from the manuscript, you can work through a book a chapter at a time as sessions come back from the booth.
If you are still casting, Screenplay prep gives you the script with every line assigned to its speaker and a cast list with each role's lines and words, the sheet you send to the actors. Upgrade that title to Cast assembly when the recordings start: nothing is charged at the upgrade, and within a year each chapter pays only the difference when it's produced.
When the recording and the script disagree
Actors improvise, fix a typo on the fly, skip a line or read one twice. After a chapter is generated, the studio's Recording panel lists every place a recording and the manuscript don't line up:
- a line that wasn't found in the actor's recording;
- a line where the text and the audio differ;
- something heard in a recording with no line behind it;
- a file that matched no line of the chapter at all, such as a recording attached to the wrong character.
Each item comes with a choice. Add audio for a missing line from a pickup, or edit the text and place it again; for a mismatch, use what the actor actually said or keep the line as placed; for an ad-lib, add it as a new line, leave it or cut it out; for a stray file, remove it. Nothing is silently patched. You decide every one.
Music and sound effects, when you want them
The order is speech only by default. When you want a score and sound effects under the assembled cast, unlock Music & SFX on the Sound Guide step; the price shows before you confirm, and finished chapters are regenerated to add it.
With the add-on, Martin, the AI audio director built into the studio, works on everything around the performances: the score, the effects and the pauses between lines. Ask for a longer beat before a reveal or for the rain to come in later, in plain words. It won't touch a recorded line; the actors' takes are not Martin's to change. Direct your audiobook by talking to it shows how working with Martin feels.
Who it is for
- Production houses recording multicast titles, where assembly is the part that eats the edit schedule.
- Audio-drama producers with actors recording remotely, who need a first assembled cut before the creative edit.
- Authors who recorded their book with a cast, friends, a writing group or hired actors, and want it to play as one production.
Cast assembly sits beside Narrator to full cast, which starts from a single narrator's recording and gives every other character a voice of their own. If your book has one narrator, that is the one; if it has a cast, this is. For a book with no recording at all, a regular Midsummerr production casts, voices, scores and mixes it in one click, and features shows the rest of the studio.
Price
Cast assembly costs $10 per 1,000 words, or $15 with Music & SFX. There is no title minimum and nothing is paid up front: generate one chapter, listen to it, and pay per chapter as you go.
To start, open Pro services from your account menu and choose Cast assembly, or see Pro services for everything we make for production teams.
FAQ
Do I need a recording for every character?
No, but nothing is invented without you. A character with no recording is flagged in that chapter, and their lines are voiced by AI only if you approve it for that character there. Everyone else is the actor's own take.
Can an AI voice replace one of my actors' lines?
No. A line cut from an actor's recording is never re-voiced, and Martin won't change it. Only a line the recording doesn't contain can be given an AI voice, and only when you choose that in the Recording panel.
Does the narrator count as a character?
Yes. Upload the narrator's recording the same way as any other actor's, and the narration is cut into place around the dialogue.
Can I try it on one chapter first?
Yes. There is no title minimum. Generate one chapter, listen to the assembly, and only then carry on with the rest of the book.
