The Best Free Transcription Software, With the Fine Print

Five ways to transcribe without paying, and the exact catch in each free tier.

"Free" means five very different things in transcription. One tool is free because it is open source and you run it yourself. One is free only if you already pay for an office suite. One hands you a starter allowance and then charges. One is free but makes you type every word. We checked the live limits on each so you can pick by the catch you can live with.

We publish this list and Pepys is one of the entries, so we ranked by genuine fit for a free need rather than by preference. Self-hosted Whisper is first because it is the cheapest path to unlimited transcription, full stop.

Our stake: Pepys publishes this roundup and is one of the listed tools; we have flagged our own entry as such. We ranked self-hosted OpenAI Whisper first because it is genuinely the cheapest way to transcribe unlimited audio, even though it is not our product. Competitor pricing and limits were verified on their live documentation.

How we judged

Real cost to use it (genuinely free vs. requires a paid subscription)
Setup effort (zero setup vs. command-line install)
Automatic vs. manual transcription
Timestamps and summaries free; speaker labels with any purchase
Privacy and what happens to your audio

1.
OpenAI Whisper (self-hosted)
Best for anyone comfortable running code who wants unlimited transcription at zero cost.
Whisper is OpenAI's open-source speech-to-text model, released under the permissive MIT license. Download the weights, run it on your own machine, and transcribe as much audio as your hardware allows with no usage fees. It is the only option here that is genuinely free and unlimited.
Pricing: Free and open source (MIT); you pay only for your own compute.
- No per-minute cost and no cap on how much you transcribe
- Open source under the MIT license
- Handles roughly 99 languages
- Runs fully offline, so audio never leaves your machine
- Requires technical setup (Python and the command line)
- No built-in speaker diarization out of the box
- Speed and practicality depend on your hardware
- No polished interface or formatted export
2.
PepysThat's us
Best for non-technical users who want an accurate transcript free with no setup, and speaker labels once they buy.
Pepys gives you the first 60 minutes free with no card required, and every file comes back with timestamps, AI summaries, chapters, and the ability to chat with the transcript. Speaker labels unlock with any purchase (3 credits per minute). It never trains on your audio or text, and auto-delete is optional. After the free minutes you buy credits that never expire.
Pricing: First 60 minutes free, no card; then pay-as-you-go credits from about $1.25/hr for small top-ups down to $1.00/hr at the recommended 20-hour tier and $0.85/hr at volume, and credits never expire (pricing as of 2026).
- 60 free minutes with no credit card
- Timestamps, AI summaries, and chapters on every file; speaker labels with any purchase
- Chat with the transcript on any file
- 99+ languages
- Never trains on your audio or text; optional auto-delete
- The free allowance is a one-time 60 minutes, then it is paid
- Paid usage is credit-based rather than a free forever tier
3.
Google NotebookLM
Best for researchers who want to question and summarize audio rather than export a clean transcript.
NotebookLM is free and lets you upload an audio file as a source, then ask questions and generate summaries grounded in it. It is a research assistant more than a transcription tool: it does not hand you a formatted, speaker-labeled transcript to export.
Pricing: Free tier; paid plans raise the caps.
- Free to use
- Accepts audio as a source you can query and summarize
- Up to 50 sources, with a generous 200MB / 500,000 words per source
- Strong for synthesizing across many documents
- Built for chat and summaries, not clean transcript export
- Free tier caps daily use (about 50 chat queries and 3 audio generations per day)
- No speaker diarization or timestamped transcript view
4.
Microsoft Word Transcribe
Best for people who already pay for Microsoft 365 and want transcription inside Word.
Word's Transcribe feature turns an uploaded or recorded file into a speaker-separated transcript you can edit right in the document. It is convenient if you are already a 365 subscriber, but there is no standalone free version and the monthly ceiling is low.
Pricing: No standalone free tier; requires an active Microsoft 365 subscription (as of 2026). Standard subscribers are capped at 300 minutes of uploaded audio per month.
- Included at no extra cost if you already pay for Microsoft 365
- Separates speakers and lets you edit in Word
- Upload a file or record directly
- Timestamps in the transcript pane
- Requires a paid Microsoft 365 subscription, so not free on its own
- Capped at 300 minutes of uploaded audio per month for standard subscribers
- Runs in Word for the web, not fully offline
5.
oTranscribe
Best for journalists and researchers who want accurate manual transcription with full privacy.
oTranscribe is a free, open-source web app that makes manual transcription less painful: pause, rewind, and change playback speed without leaving the keyboard, with your text saved locally as you type. It has no automatic speech recognition, so you write every word yourself, but nothing ever leaves your browser.
Pricing: Completely free and open source (MIT).
- Completely free with no account and nothing to install
- Audio and transcript never leave your browser
- Keyboard shortcuts and variable playback speed
- Exports to Markdown, plain text, and Google Docs
- Manual only: you type every word, with no automatic transcription
- No speaker labels or timestamps generated for you
- No AI summaries or search across the transcript

Try the pay-once option yourself

Drop in a recording for a speaker-labeled, timestamped draft. 60 minutes free, no card, and we never train on your audio.

or paste a link

More roundups

Frequently asked questions

What is the best genuinely free transcription software?

Self-hosted OpenAI Whisper is the only option here that is free and unlimited: it is open source under the MIT license, runs on your own hardware at no per-minute cost, and covers about 99 languages. The trade-off is that you install and run it yourself from the command line.

Which free tool needs the least setup?

Pepys and NotebookLM both work in the browser with no install. Pepys gives you 60 minutes free with no card, with speaker labels once you buy any pack; NotebookLM is free but is built for chat and summaries rather than exporting a clean transcript.

Is Microsoft Word Transcribe free?

No. It has no standalone free version. It is included with a paid Microsoft 365 subscription, and standard subscribers can transcribe up to 300 minutes of uploaded audio per month.

Does any free tool give me speaker labels?

Pepys labels speakers once you buy any pack (3 credits per minute), not within the free 60 minutes, and Word Transcribe separates speakers if you have a 365 subscription. Self-hosted Whisper does not do diarization out of the box, and oTranscribe is manual, so any labels are ones you type.

Why isn't oTranscribe higher on the list?

oTranscribe is excellent and free, but it does no automatic transcription at all: you type every word while it handles playback. That makes it great for accuracy-critical manual work but slow compared with the automatic options if you just want a fast draft transcript.