Tutorial 4: From document to published audiobook, with voices, Read Aloud and the Audio Studio

Goal: hear your document in the voice you want. Then use the Audio Studio to turn a folder of documents into an audiobook with chapters, or join a folder of recordings into one book, fix its chapters by ear, and publish it. Last, make a DAISY book for accessible book players.

You can do all of this from the keyboard, QUILL tells you what is happening at each step, and nothing changes until you have checked it.

1. Pick a voice

Tools > Speech > Speech and Dictation... opens the speech settings. You can choose from old and new voices:

• SAPI 5: the Windows voices you already have. Nothing to download.
• eSpeak NG and DECtalk: the classic voices, which you can download when you want them (yes, Perfect Paul is there).
• Kokoro: natural-sounding voices that run on your own computer (about a 114 MB download, once; no account needed). The QUILL Cast podcast is narrated by two of these.
• ElevenLabs (optional): a paid online voice, if you have an account. QUILL asks you before it sends anything.
• Read in Browser (experimental, Settings > Experimental): reads your document in a web page using your browser's voices, including Edge's online natural voices. QUILL tells you when a voice works online, because then your text is sent to the company that makes the voice.

Every download shows its progress and is checked to make sure it arrived safely.

2. Read aloud

Select text (or just place the caret) and use Tools > Reading & Dictation > Read Aloud. Press a key to stop. Try the same paragraph with two or three voices and keep the one you like best.

To fine-tune how a passage is read, the SSML Builder lets you add emphasis, pauses and changes in pitch and speed using ordinary controls, without typing any tags. It works with SAPI 5 and eSpeak NG.

To make a quick audio file, open the palette and choose Generate Speech Audio. It turns the current document into audio in the background. WAV always works; with the ffmpeg helper installed (Help > Download Optional Components), you also get MP3, M4A, M4B, OGG, Opus, FLAC.

3. Open the Audio Studio and choose what to do

For anything bigger than one file, Tools > Speech > Audio Studio... is the place to go (it is also in the command palette, and on Ctrl+Shift+Grave, Y). The first page asks what you want to do:

• Narrate documents: QUILL reads your files aloud into audio.
• Combine audio files: you have already recorded, and QUILL builds the book.
• Edit an existing audiobook: open an MP3 or M4B that has chapters and change them.

The wizard remembers what you chose last time and starts there. Every page is announced ("Step 2 of 7: What should I read?"), Back and Next move between steps, and Skip to summary jumps to the end when a saved project already fills in every page, so doing it again takes three keystrokes.

4. Narrate documents

Point What should I read? at a folder of documents (Word, Markdown, HTML, text). You can choose which files to include and whether to look in subfolders. Count documents tells you how many it found.

On Who should read it?:

• Pick the engine and voice, and use Preview voice to hear it first.
• Round-robin voices takes turns through a list of voices you choose: article 1 gets voice 1, article 2 gets voice 2, and so on.
• Voice casting goes further: add a rule like *interview* = your guest voice, or #1 = your narrator, and every matching chapter is read by that voice. First matching rule wins; everything else takes turns as above. This is handy for fiction with a lot of dialogue.

On the output step, two boxes are worth knowing about:

• Audition converts only the first document, so you can check the voice and pace in a few minutes before a long run.
• Reuse unchanged audio from the last run (on by default) saves you time. If you edit one chapter and run it again, only that chapter is read again; the rest say "Reused ... (unchanged since last run)". You can draft with a fast voice, then switch to your favourite voice for the final version and everything is rebuilt for you.

On the book page, type a half-remembered title and press Look up book details. QUILL looks it up in Open Library and MusicBrainz (free, and QUILL asks before it goes online the first time), fills in the author, genre and year from the match you pick, and offers to download the cover as cover.jpg. Check the spoken credits if you want the book to introduce itself.

Read the summary. If you like, choose Save a job file to keep all your choices in a .quilljob file you can load next time (or edit in Notepad). Then press Start. QUILL tells you how it is going, file by file. You can cancel at any time, or tuck the progress into the status bar.

5. Combine a folder of recordings

Recorded a memoir chapter per session? Choose Combine audio files and point it at the folder. Each file becomes a chapter named after the file, and you always get to check the chapter list before they are joined. You can rename, reorder or remove chapters, or bring in titles from a CSV file or Audacity labels. You get an M4B or MP3 with chapters, and you can also trim silence, add fades, change the speed, or match the ACX loudness rules.

Two extras for bigger jobs: Library mode builds every subfolder as its own audiobook, named after its folder, without you having to watch; and the watch action ("Build audiobook from the folder") rebuilds a watched folder's master automatically whenever new recordings land.

6. The Chapter Workbench

Open any MP3 or M4B with chapters. A long recording with no chapters opens as one big chapter, ready for you to split up.

It has a built-in player: Play and Pause, Previous and Next chapter, Rewind and Forward, a position slider that reads out the time in plain words, a speed control that does not change the pitch, and Where am I? to hear where you are. The player remembers where you stopped in every book. (Want gapless audio and exact seeking? Download the mpv player engine from Help > Download Optional Components and QUILL uses it automatically. Without it, QUILL uses its built-in player.)

To fix chapters by ear:

1. Play until you hear where a boundary belongs.
2. Choose Split at playhead to start a new chapter there.
3. Set start to playhead retimes an existing boundary; Merge into previous joins a chapter to the one before it, and Restore original undoes your changes.

Two helpers make suggestions, and you decide whether to keep them. Propose chapters from silences... listens for pauses and suggests where chapters could start. Propose AI titles... writes out the first minute of each chapter on your own computer, sends only that text to the AI you have set up, and suggests a short title for each chapter. You can keep, rename, or choose Restore original. Check against ACX tells you in plain words whether the loudness meets Audible's rules.

An MP3 is saved in place: only the chapter information changes, not the sound. An M4B is saved as a new file, with no loss of quality.

The Workbench player also has a Mute button. The separate Audio Studio app goes further: your media keys work, each book remembers its own volume, there is a sleep timer, and the next book starts by itself when one finishes. Your library there is a tree (Favorites, In Progress, Recently Played, Inbox, plus your own folders), and it can pick up the last book where you left off.

7. Publish it

The Workbench's Publish... button gives you these choices, and nothing is sent anywhere until you say so:

• Podcast feed: writes a podcast feed file (.rss) next to the book, without going online, with chapters that podcast apps can read.
• Folder feed (all episodes): run a whole show from one folder. Each finished file becomes an episode with its own description, date, length and chapters. One button rebuilds the feed after each new episode; another writes an accessible show-notes.html page beside it.
• SFTP upload: sends the book and its files to your own web server, safely. Your password is kept in Windows Credential Manager. QUILL tells you the percent as it goes, and Cancel stops it.
• Auphonic: makes your audio sound professional, using your own Auphonic account. Check account and load presets tells you how many credits you have and lists your presets. Then QUILL uploads the book, waits, and saves the results next to it. Your Auphonic key is stored safely, and you can manage it in AI > AI Hub > Services.

Build the episode, update the feed, and upload both. You can run your whole show from QUILL.

8. DAISY talking books

File > Export > DAISY Talking Book writes a DAISY 2.02 text-only talking book, the format used by accessible book players and libraries. Your headings become the way readers move around the book, so make sure your headings are in good order first (GLOW can check them for you: tutorial 6).

9. Round trip: audio in

It works the other way too. Ctrl+F9 lets you dictate without going online, and transcription turns a recording into a document on your own computer. Give a transcript to the Listening Companion and it can write meeting minutes, a to-do list or a summary for you to check. To hear more, QUILL Cast episode 21 covers Read Aloud and voices, and episode 33 covers the Audio Studio.

Next: Start an Accessible Vault.