For qualitative research
Transcribe research interviews. Your recordings and transcripts stay on your device.
Timestamped transcripts for your dissertation, thesis or study, ready for NVivo and MAXQDA. Speech recognition runs in your browser, so your participants' recordings are not uploaded to a transcription service.
60 free minutes a month · no account · Chrome or Edge on a laptop or desktop
Built for your workflow
From recording to coded data.
Everything you need between the interview and your analysis, in one place and on your own computer.
Timestamps you can click
Every sentence has a timestamp. Click it to hear that exact moment and fix misheard words right in the text. Undo if you change your mind.
De-identify in one step
Replace a participant's name, employer or town with a placeholder like P03 across the entire transcript.
Export to NVivo and MAXQDA
MAXQDA reads the timestamps and links your text to the recording when you import both. NVivo gets a table with timespan, speaker and content.
Word with line numbers, clean verbatim
A real .docx file with line numbers, so you can cite passages as “Interview 3, lines 112 to 118.” Optionally in clean verbatim style with I and P labels, [pause] markers and a timestamp at the end of each paragraph.
Several interviews at once
Drop all the recordings from your study at once. They run one after another, overnight if you like.
Key statements with quotes Pro
A short summary and the participant's most important statements, each with a verbatim quote and timestamp. Every quote is checked against the transcript, and anything not found there is left out. A starting point for your analysis, not a replacement for it.
Technical terms and names
Enter terms from your field and the names in your study before you start, so speech recognition is more likely to spell them correctly.
Who is speaking Beta in Pro
Interviewer and participant are separated automatically. You name them, and every export shows the name before each paragraph.
IRB and data management
What you can accurately say.
We can't tell you what your IRB or ethics committee will approve, and Stillscript is not certified for any specific regulation. What we can do is describe exactly how it handles your data, so you can decide with your board.
- Audio is processed on your computerSpeech recognition runs in your browser. No recording is sent to Stillscript or to any other transcription service.
- Transcripts stay in your browserThey are stored locally on your device until you delete them. You can also download a backup file and keep it wherever your data plan says.
- One download, no audioThe first time, your browser downloads the speech recognition model once. That download contains no audio and no text from you.
- Check it yourselfTranscribe one recording, then turn off your Wi-Fi. The next transcription still works.
How it works
Five steps, all on your laptop.
- Record with consentUse your phone or voice recorder, or record right in Stillscript. MP3, M4A, WAV and MP4 all work.
- Transcribe locallyDrop in the file. How long it takes depends heavily on your device: it is usually fastest with a graphics card that the browser can use. On laptops without a dedicated graphics card it can take much longer, possibly longer than the recording itself.
- Correct and de-identifyClick timestamps to check passages, fix errors and replace names with placeholders.
- Export for your analysisNVivo, MAXQDA, Word with line numbers, plain text or subtitles.
- Delete when you're doneRemove the transcript and recording from your browser whenever your data plan requires it.
Pricing
Enough for a whole study.
Start with 60 free minutes a month. For a thesis or dissertation, the Pro Pass gives you 3 months without a minute limit for a one-time $24. For research over many months, there's a subscription.
Which languages work?
You can select English or German directly; both have been tested. Other languages may work but have not been tested.
How accurate is it?
Accuracy depends on audio quality. “Best quality” works best with clear speech. Errors mostly come from strong accents, crosstalk, echo or a poor microphone. Plan time to check the transcript, as you would with any method.
Does it work on a phone?
Not reliably yet. Please use a laptop or desktop with Chrome or Edge.
Is speaker separation reliable?
It's a beta. It works best with two speakers who don't talk over each other. You can switch a paragraph to the other speaker with one click.
Read our guide to transcribing research interviews privately