Dictation apps for writing a book: what to look for and how to choose

Dictating a book means talking for an hour at a time about people and terms no speech engine has heard of, into a document you will edit for months. Every product fact below comes from the maker's own help, pricing or privacy page, checked on October 21, 2026. We name products as examples, not recommendations, and do not rank them.

What a book asks of a dictation tool

• Long sessions. Find out what happens when you pause to think, and whether there is a weekly or monthly limit.
• Punctuation and paragraphs. Can you say "comma" and "new paragraph", or does the tool add punctuation for you, and can you turn that off?
• Offline or cloud. Some tools process your voice on the device; others send it to the maker's servers. This affects privacy, and whether you can dictate on a plane or a trail.
• Privacy of client stories. Know whether your audio is stored, reviewed by people, or used to improve the maker's models.
• Getting text into a document. Text that appears where your cursor is (in Word, Google Docs or Pages) is easiest; text trapped inside an app needs copying out.
• Custom vocabulary. A tool that learns your company name, your framework and the names in your stories saves a lot of fixing.
• Cost. Built-in tools come with what you already own; dedicated tools range from limited free plans to subscriptions and one-time licenses.

Kind one: dictation built into your system or office software

These need no new account. The text goes wherever you are typing.

Apple Dictation (Mac and iPhone). Apple's Mac User Guide says you can dictate text of any length without a timeout, and that dictation stops by itself when no speech is detected for 30 seconds. You say punctuation by name, and "new paragraph" starts a new paragraph. In supported languages it inserts commas, periods and question marks automatically, a setting called Auto-punctuation that you can turn off. Keyboard settings tell you whether your dictation is processed on the Mac or sent to Apple's servers. On iPhone, Apple's iPhone User Guide says dictation is processed on the phone in many languages, with no internet connection required, and it also stops after 30 seconds of silence.

Windows voice typing. Press the Windows key and H. Microsoft's help page says voice typing uses online speech recognition powered by Azure Speech services and needs an internet connection. Its settings include automatic punctuation, and there are spoken commands for punctuation and editing. On Copilot+ PCs, a feature called fluid dictation uses on-device models to correct grammar, punctuation and filler words as you speak; it is on by default, and saying "undo that" restores what you actually said.

Google Docs voice typing. In Docs, choose Tools, then Voice typing. Google's Docs Editors Help says it works in Chrome, Edge and Safari, and that your web browser controls the speech-to-text service. Spoken punctuation (period, comma, question mark, new line, new paragraph) and editing commands work only in English,.

Microsoft Word Dictate. In Word, choose Home, then Dictate. Microsoft's Word help page says it is available only to Microsoft 365 subscribers and needs a microphone and a reliable internet connection. Spoken commands include punctuation, "delete that", "bold last sentence", and "add comment", which drops a comment into the document. Settings include auto punctuation.

Kind two: dedicated dictation software

These add features the built-in tools do not list, such as custom vocabularies, for a price or a limited free plan.

Dragon Professional v16. Nuance, now part of Microsoft, sells Dragon Professional v16 as locally installed software for Windows 11, compatible with Windows 10. Its product page describes both live dictation and transcription of existing audio files from a single speaker, custom voice commands for inserting standard text, and shared custom words for teams. Nuance states it offers "up to 99% recognition accuracy, right from first use"; that is the maker's claim, not our test. The product page shows a "Contact us" button rather than a price.

Wispr Flow. According to its pricing page, Wispr Flow works in any app on Mac, Windows, iPhone and Android, and learns your names and jargon through a dictionary. The free plan allows 2,000 words a week on desktop and limited weekly use on mobile. Pro, with unlimited dictation, is $15 per user a month billed monthly, or $12 a month billed annually. Its data controls page says transcription always happens in the cloud.

Superwhisper. Superwhisper runs on Mac, Windows, iPhone and Android. Its plans page says the free plan gives unlimited dictation with local Whisper models, custom vocabulary and text replacements, and that on the free plan nothing you dictate leaves your device. Pro is $8.49 a month, $84.99 a year or $249.99 once, and adds cloud models, AI modes that reformat your words, and file transcription. Its home page says Intel Macs work best with cloud models. For a book, a plain voice-to-text mode keeps your own sentences; an AI mode that rewrites tone can smooth your voice away.

Kind three: record first, transcribe later

This suits people who think better walking than watching words appear, and leaves you an audio master to check quotes against.

Apple Voice Memos. Apple's guide to Voice Memos transcriptions says transcription is available on iPhone 12 or later in English and several other languages, live while recording or afterwards. Choose Copy Transcript and paste it into your manuscript.

Microsoft Word Transcribe. Under Home, Dictate, Transcribe, Word can record or take an uploaded file (.wav, .mp4, .m4a or .mp3), separate speakers, and save the transcript as a Word document. Microsoft's Transcribe help page lists 300 minutes of uploaded audio a month for Microsoft 365 subscribers, says recordings are saved in a Transcribed Files folder in OneDrive, and notes that availability depends on your version of Word and type of account.

Dragon Professional and Superwhisper Pro can also transcribe audio files. For longer interviews and dedicated services, see the transcription options guide.

Side by side: what each maker says

How to choose

• You already write in Word or Google Docs: try its own dictation first.
• Your book is full of names, products and jargon: look for a tool whose maker lists a custom vocabulary or dictionary, then test it with your ten hardest words.
• Your stories are confidential: favor processing on your own device, and read the privacy checklist below before your first session.
• You think best on your feet: record voice memos and transcribe afterwards, rather than watching a screen.
• You switch between many apps: consider a dedicated tool that works in any app.
• You are not sure: run a 20-minute trial. Dictate the same 300-word passage, with your real names and terms, into two tools and count the fixes each needs.

How to dictate a chapter

1. Put the outline in front of you

   Print the chapter's five to eight points, with the story under each, or keep them on a second screen. The worksheet gives you a chapter-by-chapter plan to work from.

2. Speak in paragraphs

   One point at a time: state it, tell the story, say what it means, then say "new paragraph". Pause at the end of a thought, not in the middle of a sentence.

3. Say punctuation, or fix it later

   Decide before you start; saying every comma breaks some people's train of thought. If automatic punctuation is on, say only "new paragraph" and fix the rest when you edit. If it is off, say periods and question marks at least.

4. Flag, do not fix

   When a word comes out wrong, keep talking. In Word you can say "add comment" with a note; elsewhere, say a marker word you will search for later, such as "checkpoint".

5. One chapter per session

   Stop when the chapter's points are covered, even with energy left, and save the file with the chapter number in its name.

6. Read it back within a day

   Fix names and numbers while you remember what you meant. The transcript to chapter guide shows how to turn spoken text into prose.

Privacy checklist before you dictate client stories

Answer four questions from the maker's privacy page before you say a client's name into any tool:

1. Is my voice processed on my device, or sent to a server?
2. Is the audio or the text stored, and for how long?
3. Is it used to train or improve the maker's models, and can I opt out?
4. Can people at the company or its contractors review it?

Here is what the makers said on October 21, 2026:

• Apple. Its Siri, Dictation and Privacy page (dated September 14, 2026) says dictation not processed on the device is sent to Apple's servers and is not stored unless you opt in to Improve Siri and Dictation. It also says transcripts and related request data, tied to a random identifier rather than your Apple Account, may be kept for up to two years to improve Siri, Dictation and other features.
• Microsoft. The Microsoft Privacy Statement (last updated September 2026) says Microsoft will not store, sample or listen to your voice recordings without your permission, and that contributing voice clips to improve voice typing is your choice. Word's Dictate help page says the service does not store your audio or transcribed text. For Word Transcribe, your recordings stay in your OneDrive folder until you delete them.
• Google Docs. Google's help page says your browser controls the speech-to-text service, so read your browser maker's terms as well.
• Nuance (Dragon). Its privacy statement (updated March 14, 2025) says snippets of voice data may be manually reviewed and transcribed by employees or vendors to tune and improve its products, with steps such as de-identification where possible.
• Wispr Flow. Its data controls page (updated August 18, 2026) says your audio, transcripts and edits may be used to train its models if you allow it, controlled in the app under Settings, then Data and Privacy, and that its third-party AI providers cannot use your data for training.
• Superwhisper. Its sensitive data page says a local voice model with no AI processing keeps every stage on your machine, and that its cloud models run under zero-data-retention agreements and are not used to train AI models.

Then apply your own rules. If you are bound by client confidentiality, treat dictation like any other place client information goes. The simplest protection costs nothing: change names and identifying details as you speak, so the real ones never reach any tool. The transcription options guide covers permissions and deleting files when you are done.