Dictate
Live speech recognition that inserts text at your cursor as you speak. A useful starting point for a paragraph, email, or first draft.
Transcribe
A recording-based workflow that creates a separate, timestamped transcript for review. Better suited to interviews, meetings, and existing audio.
01 Before you start

What You Need to Use Speech to Text in Word

You need a microphone for live speech, an internet connection, and access to the relevant Microsoft 365 feature. Dictate captures speech as you talk; Transcribe processes recordings and creates a separate transcript for review.

Word speech-to-text access checklist
RequirementWhat to check
Microsoft 365 access Plan on a paid Microsoft 365 subscription for desktop Dictate. A one-time purchase of Word does not automatically include the same cloud speech features. Transcribe requires an eligible account.
Supported Word version Dictate is available in supported Windows, Mac, and web editions. Transcribe availability differs by platform and subscription.
Compatible browser For Word's web-based Transcribe workflow, use a current version of Microsoft Edge or Google Chrome.
Microphone permission Your operating system and, when relevant, browser must allow microphone access for live dictation or recording.
Internet connection Word's built-in speech services process audio online.

For the most straightforward Transcribe instructions below, use Word for the web. Supported Windows desktop editions also provide the feature. If it is absent from your Mac desktop app, check the web version rather than searching through unrelated ribbon settings.

Dictate vs. Transcribe in Word.

Dictate
  1. Live speech
  2. Home > Dictate
  3. Text appears in your document.
Transcribe
  1. Audio file or recording
  2. Home > Dictate dropdown > Transcribe
  3. Review transcript
  4. Add to document.
02 Speak into your document

How to Use Live Dictation in Word

Place the cursor where text should appear, select Home > Dictate, allow microphone access, and start speaking. Word inserts text as you talk; select the microphone button again to stop.

  1. Open your document and position the cursor

    Create a blank document or open an existing one. Click exactly where you want the dictated text to go.

    Word inserts speech at the active cursor position. If you click into an earlier paragraph while dictation is running, new text can land there instead.

  2. Select Home > Dictate

    Open the Home tab and find Dictate, usually represented by a microphone icon.

    Select the main button for live speech recognition. The adjacent dropdown, where available, contains Transcribe.

  3. Allow microphone access

    Accept the microphone permission prompt.

    If you use a headset, check that Word is receiving audio from that microphone rather than a distant laptop microphone. A close microphone generally captures clearer speech with less room noise.

  4. Check the spoken language

    Open the dictation settings and select the language you are speaking.

    This matters. Dictating Spanish while English recognition is selected is not a microphone problem, even if the result looks like one.

  5. Speak naturally, then review

    Speak at a steady pace and pause briefly between ideas. You do not need to stretch every syllable.

    When you finish, stop dictation and check names, numbers, acronyms, and technical terms. Recognition errors often produce real words, so a spelling check alone will not catch them.

The awkward part: punctuation and formatting commands

With automatic punctuation switched off, speaking a sentence does not reliably produce a correctly punctuated sentence. You need commands such as "period" and "comma."

Common spoken commands in Word
Say thisWord inserts or performs
"Period"A full stop
"Comma"A comma
"Question mark"A question mark
"New paragraph"A paragraph break
A spoken example with manual commands

"Hi Alex comma the report is ready period new paragraph please review page three period."

It works. It also makes a simple voice note feel like operating a document by remote control. Remembering punctuation commands can interrupt your train of thought, especially during longer dictation.

Word does offer automatic punctuation in supported versions and languages. Open the Dictate settings and look for Enable auto punctuation. Turning it on reduces the need to speak punctuation marks, although precise paragraph breaks and formatting still require commands or keyboard edits.

The distinction matters: Word does not always require spoken punctuation. But its manual-command workflow becomes cumbersome when automatic punctuation is unavailable, disabled, or not matching your intended sentence structure.

03 Work from a recording

How to Transcribe Audio Files in Word

Open Home > Dictate dropdown > Transcribe and choose Upload audio to transcribe an existing recording. Review the transcript, correct speaker labels, then select Add to document.

Upload an existing recording

  1. Open Word and sign in. Use an account with access to Transcribe.
  2. Open the Transcribe pane. Select Home > the arrow beside Dictate > Transcribe.
  3. Select Upload audio. Choose the recording from your device.
  4. Wait for processing. Leave the Transcribe pane open while Word processes the file. Processing time depends on the recording and connection.
  5. Review the transcript. Use the timestamps and playback controls to check unclear passages.
  6. Correct speaker labels. Replace generic labels such as "Speaker 1" with names where appropriate.
  7. Select Add to document. Insert the transcript using the available speaker and timestamp options.

Microsoft's listed upload formats: WAV MP4 M4A MP3

An MP4 can contain video, but Transcribe processes the speech. It does not describe slides, demonstrations, or other visual content.

Speaker labels also need checking. Two people talking over each other can produce incorrect attribution, even when individual words are recognizable.

Record directly inside Word

You can also capture a new recording through the Transcribe pane:

  1. Select Start recording.
  2. Allow microphone access.
  3. Record the conversation, using pause and resume as needed.
  4. Select Save and transcribe now when finished.

This is not the same as Dictate. Dictate inserts text while you speak. Transcribe creates a recording and generates a transcript for review after processing.

For meetings or interviews, get the participants' consent before recording where required.

04 Plan for the practical limits

Word's Transcription Limits: Subscriptions, Files, and History

Word's main transcription constraints are subscription eligibility, uploaded-audio allowances, file restrictions, and document-based transcript storage. These matter most when you record frequently or process long interviews.

Uploaded audio has a monthly allowance

Microsoft's Transcribe support guidance lists a standard allowance of 300 minutes of uploaded audio per month for Microsoft 365 subscribers. That is 5 hours. Some eligible licenses provide higher allowances, so check the quota attached to your account.

Standard uploaded-audio allowance

300 min / month
5 hours

Example: each interview lasts 75 minutes. The bars show cumulative use of a 300-minute allowance.

1 interview 75 minutes total
25%
2 interviews 150 minutes total
50%
3 interviews 225 minutes total
75%
4 interviews 300 minutes total
100%
Four 75-minute uploads use the full standard allowance. This is an illustrative workload, not an account usage report; eligible licenses can have different quotas.

The important detail is uploaded audio. Do not assume every recording workflow or subscription has identical limits.

File-size restrictions complicate large uploads

Large recordings can run into upload restrictions before transcription starts. If Word reports that a file is too large, create a smaller supported copy or split the recording into numbered sections.

Keep the original.

For longer projects, check the current upload-size restrictions and your remaining transcription allowance before preparing the entire batch.

Why renaming a file does not convert it

Changing a filename from .mov to .mp3 does not convert the media. Use an actual conversion tool if the source format is unsupported.

Repeatedly compressing an already compressed recording can also damage speech clarity. Work from the original recording when making a smaller supported copy.

Transcript history is tied to documents and OneDrive

Word stores Transcribe recordings in OneDrive's Transcribed Files folder. The transcript is associated with its Word document rather than collected in a dedicated voice-note library.

There is another catch: Word supports one active transcript per document. Starting a new transcription replaces the previous transcript in that document's pane, although text already inserted into the document remains.

That creates extra housekeeping:

  • Create separate documents for separate recording sessions.
  • Use consistent names for documents and audio files.
  • Keep track of which OneDrive recording belongs to which document.

Word has document history and file management. What it lacks is a focused voice-note history that brings those tasks together. For occasional transcription, this is manageable. For daily voice notes, it gets old fast.

Where Word fits

  • Draft text directly into the document you are editing.
  • Keep final formatting and document work in one place.
  • Handle occasional recordings with eligible access.

Where it adds work

  • Uploaded audio is subject to account allowances.
  • Each document holds one active transcript.
  • Finding voice notes means managing documents and stored recordings.
05 A speech-first alternative

Better Alternatives: SpeechText.AI for Recordings and Live Dictation

Use SpeechText.AI for audio and video transcription, and its AI Voice Note Taker for live, browser-based dictation. One workflow processes existing recordings; the other turns spoken thoughts into punctuated notes with a dedicated file history.

SpeechText.AI for audio and video transcription

If your starting point is a collection of interviews, lectures, meetings, or recorded presentations, use a transcription-first workflow rather than treating each recording as a Word document.

SpeechText.AI provides audio and video transcription with domain-specific models and multi-channel audio processing. Those features address practical recognition problems that basic dictation does not solve by itself.

Domain-specific models help with specialist vocabulary. A financial discussion, medical lecture, and general conversation do not use the same terminology. Selecting an appropriate model gives the recognition system relevant language context. You should still check names, figures, abbreviations, and consequential statements.

Multi-channel processing preserves useful recording structure. If an interviewer and guest were recorded on separate channels, processing those channels separately supports clearer speaker handling.

Separate channels are not the same as a mixed recording

Processing separate audio channels is different from identifying several speakers in a single mixed track. Separate channels carry information that a mono recording no longer contains.

If everyone was recorded into one mixed track, multi-channel processing cannot recover separate original channels that were never saved.

A practical workflow is:

  1. Upload a supported audio or video file.
  2. Select the language and relevant processing options.
  3. Generate the transcript.
  4. Review uncertain wording against the recording.
  5. Copy the finished text into Word for editing, formatting, or sharing.

Keep Word for the document work. Let SpeechText.AI handle the transcription work.

Before submitting a large project, check current pricing and upload allowances. No transcription service removes the need to plan around recording quality and workload.

AI Voice Note Taker for natural, browser-based dictation

For live speech, use the SpeechText.AI AI Voice Note Taker. Open it in your browser, allow microphone access, and start a voice note without making a word processor the center of the process.

  • Automatic punctuation by default. Speak naturally while the tool adds punctuation. You do not need to build spoken "period" and "comma" commands into your normal note-taking routine.
  • Dedicated file history. Return to earlier notes and transcripts without matching separate Word documents to recordings in OneDrive.
  • Browser-based access without a desktop application. Start from the browser without installing or updating a desktop office suite.

Automatic punctuation still deserves a quick review, especially for questions, lists, names, and numbers. The benefit is less cleanup and fewer interruptions, not a promise that every sentence needs zero editing.

Word vs. SpeechText.AI: which workflow fits?

Choose by task, not just by the microphone button
Requirement Microsoft Word SpeechText.AI
Live speech capture Dictate inserts text into an open document. AI Voice Note Taker provides a dedicated voice-note workflow.
Automatic punctuation Available in supported versions and languages; check Dictate settings. Enabled by default in the Voice Note Taker.
Voice-note history Managed through documents and stored recordings. Dedicated file history in the Voice Note Taker.
Existing audio and video Transcribe supports specified formats and account allowances. Dedicated audio/video transcription with domain-specific models and multi-channel processing.
Installation Desktop Word requires installation; Word for the web does not. Browser-based access without a desktop application.
Final document formatting Full document-editing tools. Copy the reviewed transcript into Word when needed.

Word also works in a browser, so browser access alone is not the distinction. The stronger reason to choose SpeechText.AI is the focused workflow: capture speech, find previous notes, review transcripts, and move the text wherever it belongs.

Document-first vs. voice-note-first.

Microsoft Word

  1. Open document
  2. Dictate or Transcribe
  3. Manage document and recording.

SpeechText.AI Voice Note Taker

  1. Open browser tool
  2. Speak naturally
  3. Review punctuated note
  4. Find it in file history.
06 Get it working

Fix Common Word Speech-to-Text Problems

Check account eligibility, microphone permissions, Word support, and speech-language settings first. For Transcribe failures, also check the format, file size, remaining quota, and OneDrive access before reinstalling anything.

Match the symptom to the first checks
ProblemWhat to check
Dictate is missing Confirm that you are signed into the correct Microsoft 365 account, using a supported edition, and permitted to use connected experiences. Update Word if necessary.
The microphone does not respond Check operating-system permissions, browser site permissions, and the selected input device. Disconnect unused microphones if the wrong one is active.
Transcribe is missing Open the dropdown beside Dictate. Try Word for the web in Edge or Chrome and confirm your account's eligibility. Work or school administrators can restrict features.
Words appear in the wrong language Set the spoken language in Dictate settings. The document's proofing language is not a substitute for the speech-recognition setting.
Punctuation is missing Check automatic punctuation in Dictate settings, or speak punctuation commands. Support varies by language and version.
An audio upload fails Confirm the file is a genuine supported format, check its size and your remaining quota, and verify OneDrive access. Convert unsupported media rather than renaming its extension.

Can you dictate in Word without Microsoft 365?

You can use operating-system speech input to enter text into a document without using Word's own Dictate service.

  • Windows: Click in a text field and press Windows + H to open voice typing.
  • Mac: Turn on System Settings > Keyboard > Dictation, then use the configured shortcut.

These are separate tools. They do not provide Word's Transcribe workflow for existing recordings, and you still need access to a document editor.

For a dedicated browser-based alternative with voice-note history, use SpeechText.AI's AI Voice Note Taker.

What if recognition accuracy is poor?

Fix the recording conditions first. Move closer to the microphone, reduce background noise, and avoid overlapping speech.

Then check the language and review specialist terms. For important recordings, test a short representative sample before processing the full file. A clear 60-second sample can reveal microphone, vocabulary, and speaker-separation problems early.

Skip the Word Frustration and Start Speaking

Use SpeechText.AI's AI Voice Note Taker for naturally spoken, automatically punctuated notes with dedicated file history. Have audio or video already recorded? Transcribe it with SpeechText.AI, review the text, and bring the finished result into Word.