Transcribe audio to text free, on your PC
No sign-up, nothing uploaded, no minute cap. The .txt comes with the prompt already on top for your AI and the .srt with the minute of every sentence.
Updated: October 2026 · 6 min read
To transcribe an audio file to text for free on Windows, with no sign-up and without uploading it anywhere: install AppGrabbit, open Download › Transcribe with AI, drop the file (mp3, m4a, wav, ogg, opus…) and click Transcribe. The AI engine runs on your computer and leaves a clean .txt, with the prompt you pick on top for your AI, and an .srt with the minute of every sentence. No minute cap. 1 credit · 5 free a month
Why “audio to text free online” is rarely free
The websites that rank first give you 10 or 30 minutes and then ask for a card, make you create an account, and keep the audio on their servers. If it is a meeting, a consultation or a class, that matters. Phone apps transcribe well but do not leave you a file you can paste into Word. And dictation in Word or Google Docs only hears what you say live: it does not take a file.
How AppGrabbit does it
You drop the file and the app turns it into a light WAV with FFmpeg and transcribes it with Whisper (whisper.cpp) on your PC. A voice detector skips silences and music, and a filter removes what these engines tend to invent where nobody is talking. Out come the .txt with the prompt on top (summarize, key ideas, study questions) and the .srt with timing, in your Transcripts folder and in the Library. That is what you paste into Word, ChatGPT or your notes.
How long it takes, in numbers
- With an NVIDIA card — about 30 seconds per 10 minutes of audio, with either model.
- CPU only — about 2 minutes per 10 minutes of audio with the Fast model; Accurate takes about 6 and makes fewer mistakes.
- Once — the engine and the model download the first time (Fast 181 MB, Accurate 547 MB; with NVIDIA the engine is 643 MB). The app states the size and the time before starting.
AppGrabbit against the other ways
| AppGrabbit | Audio-to-text websites | Word or Docs dictation | Phone apps | |
|---|---|---|---|---|
| Takes a file | ✓ Yes, drop it | ✓ Yes, by uploading | ✗ Only what you say live | ~ Some |
| Your audio never leaves your PC | ✓ Never | ✗ Goes to their servers | ✗ Goes to Microsoft or Google | ✗ Goes to their cloud |
| Sign-up | ✓ None | ✗ Account and email | ~ Your Office or Google account | ✗ Account |
| Minute cap | ✓ None | ✗ 10 to 30 free, then paid | ✓ None | ~ Monthly |
| Prompt for your AI included | ✓ Yes | ✗ No | ✗ No | ✗ No |
| Cost | 1 credit per transcript (5 free a month) · Pro $8.99/mo · 7-day pass $3.99 | Free minutes, then a subscription | Included with Office 365 or Google | Subscription |
Step by step
- 1Install AppGrabbit for Windows — Download the free installer (everything included). Windows will say it protected your PC because the installer is not code-signed: click More info, then Run anyway.
- 2Open Download › Transcribe with AI and drop the audio — mp3, m4a, aac, wav, ogg, opus, flac, wma… The “Audio or video from your PC” area reads it where it is, without copying or uploading it.
- 3Pick the model the first time — Fast (181 MB) or Accurate (547 MB); the one recommended for your PC is pre-selected. With an NVIDIA card the engine uses the GPU and takes about 30 seconds per 10 minutes of audio.
- 4Pick the prompt and click Transcribe — Summarize, Key ideas, Study questions or no prompt: it goes at the top of the .txt so you can paste it into ChatGPT, Claude or Gemini. The first time, the app gets the AI engine ready on your PC (a one-time download) and carries on by itself.
- 5Open the .txt and the .srt — They land in the Transcripts folder inside your downloads folder, and in the Library. The .txt pastes into Word or into your AI; the .srt carries the minute of every sentence.
Frequently asked questions
1 credit per transcript. Free comes with 5 credits a month and, at 0, you can claim 1 more every 48 hours. The 7-day pass (USD 3.99, no renewal) removes credits for a week; Pro removes them for good at USD 8.99 a month. No minute cap and no file-length cap.
Yes. The .txt opens and pastes into Word as is, with the prompt on top if you picked one. If you want the timing, the .srt is plain text too and opens in Notepad.
No. A three-hour audio transcribes the same; it just takes longer. Free websites cut you off at 10 or 30 minutes; here the limit is your PC.
Yes. Save it from WhatsApp Web or Desktop and drop it in. There is a guide just for that, with the step for saving the voice note.
The engine detects the spoken language. If languages mix, pin one (Spanish, English, Portuguese and more) before transcribing. It comes out in the language of the audio, not translated.
The Fast model gets about 8 words in 100 wrong; the Accurate one about 5. Mostly proper names, acronyms and strong accents; the meaning holds. That is why the .srt sits next to it: if something looks off, jump to the exact minute and fix it.
Nothing. The audio is transcribed by Whisper (whisper.cpp) running on your computer, on your NVIDIA card if you have one or on the CPU if not. You can unplug the internet and it keeps working. Only the first time does it download the engine and the model.
Related guides
Turn your audio into text, on your PC
Windows 10/11 · 1 credit per transcript, 5 free a month · No sign-up, nothing uploaded