Scrieb
Windows · Free tools · Honest comparison · Updated Sep 2026

Free Offline Transcription Software for Windows and Mac: The Real Options

Yes, you can transcribe audio on a Windows PC for free, offline, with no upload. Four free tools do it well, and Scrieb, which I make, has a free plan of 1 hour a day. Read the recommendations with that in mind; the open-source tools are listed first because they may be all you need.

All four run OpenAI's open-source Whisper model on your own computer. They differ in interface, speaker labels and how much setup they demand.

Download for MacDownload for Windows
Free: 1 hour of recording a day, no card needed.Windows on ARM64 (Snapdragon)

The four free tools

1. Buzz

Open-source desktop app. Installer, a window, drag in a file, get text. Labels speakers too; exports TXT, SRT and VTT. The right answer for most people who want free.

2. WhisperX

Command line. Adds speaker labels and word timestamps to Whisper. Needs Python, a Hugging Face token and, for GPU speed, matching CUDA. An afternoon of setup.

3. Whisper CLI

OpenAI's reference tool. Command line, Python, no speaker labels. The cleanest option if you script your workflow.

4. whisper.cpp

A fast C++ port that runs well on CPUs. Command line, some prebuilt Windows binaries. No speaker labels. Good on older PCs without a GPU.

Scrieb library listing eight transcribed and summarized recordings: meetings, interviews, a lecture and a voice memo
Scrieb’s library: recordings transcribed and summarized on the computer. The free plan covers 1 hour of transcription a day, with no account.

Side by side, including the paid option

BuzzWhisperXWhisper CLIwhisper.cppScrieb
PriceFreeFreeFreeFreeFree 1 h/day, Pro paid
InterfaceAppTerminalTerminalTerminalApp
Speaker labels✓✓✗✗✓
DOCX / PDF export✗✗✗✗✓
Local summary✗✗✗✗✓
GPU without manual setup✓✗✗✗✓
Video files✓via ffmpegvia ffmpegvia ffmpeg✓
Also on Mac✓✓✓✓✓

Which one to pick

  • Lectures, memos, and interviews where TXT or subtitles are enough: Buzz. Free and done.
  • Interviews or meetings, and you know Python: WhisperX. Free speaker labels, some setup.
  • Old PC without a graphics card: whisper.cpp, or Buzz with a small model.
  • Interviews or meetings, no terminal, need a finished document: that is the gap Scrieb fills, and the only case where I'd tell you to pay.

On a Mac

  • MacWhisper (free tier): polished Mac app; speaker labels and larger models are in the paid version.
  • Buzz: the same open-source app as on Windows, free, with speaker labels; new versions need Apple Silicon.
  • macOS dictation: free and on-device on Apple Silicon, but it types what you say live; it does not transcribe recordings.
  • Scrieb free plan: 1 hour of recording a day with speaker labels, summaries and Word export, on Mac and Windows.

What all of them need from your PC

Windows 10 or 11, 8 GB of RAM, and patience if there is no dedicated GPU: CPU-only transcription runs at roughly real time. An NVIDIA RTX 20-series or newer takes a one-hour file down to 4–12 minutes. Details in offline transcription for Windows.

Try it on your own recordings

Download for MacDownload for Windows
Free: 1 hour of recording a day, no card needed.Windows on ARM64 (Snapdragon)

Frequently asked questions

What is the best free offline transcription software for Windows?
Buzz, if you want an app with a window: free, open source, drag in a file, get text with speaker labels. WhisperX, if you are comfortable in a terminal. Whisper's own command line if you script things. All three run OpenAI's Whisper model on your PC with no upload.
Is there a free offline audio-to-text converter for Windows with speaker labels?
Yes. Buzz, a free open-source desktop app, now labels speakers, and so do the free apps Vibe, noScribe and aTrain. WhisperX does it from the command line and needs Python, CUDA drivers for GPU speed and a Hugging Face account for the speaker model. Scrieb labels speakers too and adds summaries and DOCX/PDF export; it is paid after the trial.
Do free tools need a graphics card?
No, they run on the CPU, slowly: roughly real time or slower. An NVIDIA RTX card makes every one of them 5 to 15 times faster, but getting GPU acceleration working in the command-line tools means installing the right CUDA build yourself.
Is Windows' built-in dictation an option?
Windows voice typing transcribes live speech from a microphone, not audio files, and on most systems it sends audio to Microsoft. It is not a tool for transcribing recordings.
Why would I pay for Scrieb if free tools exist?
For DOCX and PDF export, a summary written for the type of recording, a custom vocabulary, GPU acceleration that works out of the box, and one signed app for Windows and Mac. If none of that matters to you, use Buzz.
Is Scrieb free?
Partly. The free plan gives you 1 hour of recording per day with speaker labels, summaries and all export formats, no card needed. Pro removes the limit. If you need more than an hour a day and don't want to pay, the free tools above are good.

Related pages

→ Whisper for Windows: Every Way to Run It→ Offline Transcription for Windows→ Best Local Transcription Software→ MacWhisper for Windows→ Microsoft Word Transcribe Alternative→ Local Transcription