Dictation for writers: drafting a book by voice
Dictating a book is not typing with your mouth. It works when you treat the spoken draft as raw clay and the keyboard as the tool for shaping it later. Writers who stick with it usually change three things: they prepare before they speak, they never edit while dictating, and they measure their own output instead of chasing someone else's number. Here is a workflow you can start this week, with concrete routines and the small habits that make the edit bearable.
Why voice drafting works (and when it does not)
Speaking skips the habit of rereading the last sentence ten times. You keep moving because there is no blinking cursor inviting you to fix a word. That makes dictation strong for first drafts, scene sketches, interview-style nonfiction and long explanations you already understand.
It is weaker for line-level polish, tight poetry, and passages where the exact rhythm of every word matters on the first pass. Most voice writers end up with a split: talk the draft, type the revision. If you expect the spoken version to be publishable, you will be disappointed and quit after a week.
Set up once: microphone, document, one tool
- Microphone: a headset or a desk microphone at a consistent distance beats a laptop's built-in mic in a room with echo. Consistency matters more than price.
- One draft file per book: dictate into the same manuscript document (Scrivener, Word, Google Docs, a plain Markdown file). Start each session under a heading with the date so you can find today's raw text.
- One dictation tool: pick one that types into any app, so you can move between your manuscript, your notes file and email without changing habits. The overview of dictation apps compares the main choices.
- A notes file open beside the manuscript: character names, places and invented words in their correct spelling, ready to paste.
If the manuscript is under contract or embargo and you work on a shared computer, read how private dictation is before you pick a tool, because some send your voice to a server.
Run a dictation session
Before speaking, spend three to five minutes writing a scene list by hand or keyboard: who is there, what changes, the last line you want to reach. Talking without a target produces rambling; talking toward a target produces scenes.
Then dictate in short sprints. Twenty to twenty-five minutes is a good starting length because your voice stays fresh and you can count output per sprint. Between sprints, stand up, drink water, glance at the scene list. Do not read back what you said.
- Leave spoken markers instead of stopping to fix things: say “bracket check name bracket” or simply “TK” where a fact or name needs checking later. TK is the old editorial mark for material to come, and it is easy to search for.
- If you lose the thread, say the next beat out loud as a plain summary (“she leaves the house angry and drives to the coast”) and keep going. The summary becomes a placeholder you expand later.
- Keep a glass of water within reach. A dry throat slows you down faster than tired ideas.
Hands-free mode for long takes
Hold-to-talk is ideal for a sentence or two, but holding a key for a whole scene is uncomfortable. Most good tools have a toggle mode: one tap starts listening, the next tap stops. Use it for full sprints so you can lean back, gesture, pace a step away from the desk or look out the window while you talk.
Two cautions. First, stay within range of the microphone; walking across the room drops quality quickly. Second, check where the cursor is before you start. In toggle mode, text goes wherever focus is, and twenty minutes of chapter in a chat box is a bad surprise.
Dialogue, punctuation and formatting
Dialogue is the hardest part of voice drafting. You can say every mark, for example: “open quote I am not going back there comma close quote she said period new paragraph”. That is precise but slow and pulls you out of the scene.
Many writers do the opposite: speak dialogue naturally, pause at the end of each line, say “new paragraph” when the speaker changes, and add quotation marks in the edit. Try both for one sprint each and keep whichever lets you stay in the story. Chapter headings, italics and scene breaks are faster to add by keyboard afterwards.
The editing pass
Let the raw text rest at least a day. Then edit in a fixed order so each read has one job:
- Search pass: find every TK and bracket marker, fill or note them.
- Names pass: search for each recurring name and fix misheard variants. If a name keeps coming out wrong, consider spelling it differently in your notes file and pasting it, or slowing down only on that word.
- Structure pass: read for scene order and repetition. Spoken drafts repeat ideas more than typed ones, because you circle back aloud.
- Line pass: now type. Cut filler phrases you say but would never write, such as “and then” chains and “sort of”.
Expect the edited text to be shorter than the dictated text. That is normal and a sign the process is working.
Example routines and targets
Do not adopt a word count from the internet. Measure yours: after your first three sessions, divide the words in each raw draft by the minutes you spoke. That is your rate. Set a session target at roughly that rate times the minutes you have, then raise it slowly once sessions feel easy.
| Writer | Routine | Target style |
|---|---|---|
| Novelist with a day job | Two 25-minute sprints before work, scene list prepared the night before | One scene per morning |
| Nonfiction author | One 40-minute session in toggle mode explaining a chapter section as if to a student | One section outline turned into prose |
| Memoir writer | Three 15-minute sessions a week answering a single prompt, such as “the kitchen in 1994” | One remembered scene per prompt |
| Revision week | No dictation; typing only, working through the editing passes | One chapter edited per day |
Count words of finished, edited text weekly as well as raw dictated words. The gap tells you whether to plan more drafting or more editing time.
Where Voice Monkeys fits
Voice Monkeys types into any app at the cursor, so it works with Scrivener, Word or a plain editor without plugins. Hold right Option (Mac) or right Alt (Windows, Linux) for a sentence; tap once to go hands-free for a whole sprint and tap again to stop; Esc cancels a take. On Apple Silicon Macs it runs on the Mac and keeps working offline, which suits writing retreats with bad Wi-Fi. It does not transcribe audio files, so if you prefer recording on a phone while walking and transcribing later, a file transcription service is the better tool. You can download it and try a single sprint during the free trial. If you work on a Mac and want the built-in route first, the Mac dictation setup guide covers it.
Questions and answers
Can you really write a whole novel by dictation?
The first draft, yes, if you plan scenes before speaking and accept that you will revise by keyboard. Very few writers publish the dictated text without a thorough edit.
How many words per hour should I dictate?
There is no universal number. Measure your own rate over three sessions and set targets from that, raising them gradually.
Should I say punctuation out loud when dictating fiction?
Try both. Spoken punctuation is precise; natural speech with pauses and “new paragraph” keeps you in the scene. Pick the one that keeps you drafting.
What about character names the software keeps getting wrong?
Keep correct spellings in a notes file, mark doubtful spots with TK, and fix them all with search during the editing pass.
Is dictation good for writers with wrist pain?
Many writers use it to cut keyboard time. If you have pain, see a clinician as well; dictation reduces typing but does not replace medical advice.
Sources
- Apple: Use Dictation on Mac (checked November 6, 2026)