Short answer: To transcribe audio on Windows, you have three free routes — Microsoft Word's "Transcribe your recordings" (upload a file, capped at 300 minutes/month with a Microsoft 365 subscription), Windows Voice Typing (press Win+H to dictate live speech), or Audacity paired with OpenAI's free Whisper model for offline transcription. Each has real limits: caps, no live summary, or a technical setup. For a transcript with automatic speaker labels plus an AI summary and action items in 100+ languages, use an AI note taker like CraftNote, which runs on Windows, Mac, mobile, and the web.
← Back to Blog
How to Transcribe Audio on Windows (Free and AI Methods, 2026)
The Windows transcription options at a glance
"Transcribe audio" can mean two different jobs: turning an existing recording (an MP3 of an interview, a meeting file, a voice memo) into text, or converting speech into text live as you talk. Windows has free tools for both, but none of them bundle the transcript with a usable summary. Here is how the main methods compare.
| Method | Best for | Cost | Speaker labels | Key limit |
|---|---|---|---|---|
| Word "Transcribe your recordings" | Uploading a finished recording | Needs a Microsoft 365 subscription | Yes (Speaker 1, 2…) | 300 minutes/month cap; web version needs Edge or Chrome |
| Windows Voice Typing (Win+H) | Live dictation into any text box | Free, built in | No | Live speech only; needs an internet connection |
| Audacity + Whisper | Offline, unlimited-length files | Free, open source | No (not natively) | Command-line setup; manual, slow on long files |
| CraftNote (AI note taker) | Meetings, calls, and files with a summary | Free plan; paid from $9.99/mo | Yes, automatic | Recording caps on the free tier |
Method 1: Microsoft Word — "Transcribe your recordings"
If you already pay for Microsoft 365, Word can transcribe an audio file and separate the speakers for you. It is the most capable of the built-in options.
- Open Word in Microsoft Edge or Chrome (the upload feature runs in Word for the web).
- On the Home tab, click the arrow next to Dictate, then choose Transcribe.
- Select Upload audio and pick your file. Word accepts .wav, .mp3, .m4a, and .mp4.
- Wait for processing, then click Add to document to insert the transcript with timestamps.
The limits to know: Word transcription requires a Microsoft 365 subscription, and it caps uploaded audio at 300 minutes per month (users on a Microsoft Copilot license get 30,000 minutes). It needs an internet connection, the upload flow only works in Edge or Chrome, and while it labels speakers as "Speaker 1" and "Speaker 2," you still rename them by hand. Most importantly, you get a raw transcript — no summary, no action items, no way to ask questions about the conversation later.
Method 2: Windows Voice Typing (Win+H)
Every Windows 11 PC includes Voice Typing, a free live-dictation tool. It does not transcribe a saved file directly — it types whatever your microphone hears in real time.
- Click into any text field (a Word doc, Notepad, a browser box).
- Press Windows logo key + H to open the small dictation toolbar.
- Click the microphone icon and start speaking; the words appear where your cursor is.
- Open the gear icon to turn on auto punctuation so it adds commas and periods for you.
- Say "stop listening" or click the mic again to end.
The limits to know: Voice Typing is built for you talking directly into the mic, so transcribing a recording means playing it aloud and letting the PC re-listen — which hurts accuracy. It relies on an internet connection because speech is processed on Microsoft's servers, it has no concept of separate speakers, and there is no transcript file, summary, or search — just text dropped into whatever app you had open.
Method 3: Audacity + Whisper (free and offline)
For a fully free, no-subscription route that handles long files, you can pair Audacity (free, open-source audio editor) with Whisper, OpenAI's open-source speech-to-text model that runs locally on your PC.
- Use Audacity to record or clean up your audio, then export it as a WAV or MP3.
- Install Whisper on Windows (via Python, or a packaged app like Whisper Desktop that gives it a simple interface).
- Point Whisper at your exported file and let it generate a text transcript — it runs offline once the model is downloaded.
The limits to know: Whisper is genuinely accurate and has no monthly cap, but the setup is technical — the base tool is command-line, and transcribing a long recording on a laptop without a strong GPU can be slow. Out of the box it does not label who spoke, and, like the others, it stops at a plain transcript with no summary or action items.
The AI method: transcript, speakers, and a summary in one step
The free tools all leave you with the same homework: a wall of text you still have to read, clean up, and mine for decisions. An AI note taker closes that gap. CraftNote records your PC's system audio and microphone bot-free — nothing joins your Teams, Zoom, or Google Meet call as a visible participant — or transcribes a file you import, and then does the parts the built-in tools skip.
- Automatic speaker labels so you can follow who said what without renaming anything.
- 100+ languages with automatic language detection — well beyond the handful Voice Typing supports.
- An AI summary plus separated action items, so an hour of audio becomes something you skim in two minutes.
- Search across your whole archive and chat-with-your-notes to pull answers out of past recordings.
- One account across Windows, Mac, iPhone, iPad, Android, and the web, with notes that sync everywhere and record offline.
To try it, download the CraftNote Windows app, start a recording (or import an existing file), and it produces the transcript, speaker-labeled notes, summary, and action items on its own. The free plan includes 2 hours of recording per week and 10 uploads, with paid plans from $9.99/month if you need longer recordings.
Which method should you use?
- You already pay for Microsoft 365 and just need a file turned into text: Word Transcribe, as long as you stay under 300 minutes a month.
- You want to dictate notes or an email hands-free: Windows Voice Typing with Win+H.
- You need free, offline, long-form transcription and are comfortable with setup: Audacity + Whisper.
- You transcribe meetings and calls and want a summary and action items, not just text: an AI note taker like CraftNote.
Frequently Asked Questions
Can Windows transcribe audio for free?
Yes. Windows Voice Typing (Win+H) is free and built in for live dictation, and Audacity paired with OpenAI's Whisper model transcribes files offline at no cost. Word's Transcribe feature works too, but it requires a paid Microsoft 365 subscription.
Does Microsoft Word have a transcription limit?
Yes. With a Microsoft 365 subscription you can transcribe up to 300 minutes of uploaded audio per month. Accounts with a Microsoft Copilot license get up to 30,000 minutes. The file-upload feature also only runs in Microsoft Edge or Chrome.
How do I transcribe a recorded meeting on Windows?
Upload the file to Word Transcribe, or run it through Whisper for a free offline option. For automatic speaker labels plus a summary and action items, import the recording into an AI note taker like CraftNote, which handles meetings and calls end to end.
Can I get speaker labels on a Windows transcript?
Word Transcribe separates speakers generically ("Speaker 1," "Speaker 2") and you rename them yourself. Voice Typing and a default Whisper setup do not identify speakers at all. CraftNote applies speaker labels automatically.
Does transcription work offline on Windows?
Whisper runs fully offline once the model is downloaded. Word Transcribe and Windows Voice Typing both need an internet connection. CraftNote can record offline and syncs the transcript when you reconnect.
Related reading
Turn this into action with CraftNote
Record meetings and lectures, then get AI transcripts, summaries, and action items.
Try CraftNote free
