Offline audio transcription, speaker identification, and AI summaries — all on your own computer. No monthly fee. Nothing uploaded to the internet unless you choose to.
Free download · Signed installer · Works fully offline after first launch · Mac version coming soon.
VALU-SCRIBE Personal turns recordings into readable, speaker-labeled transcripts entirely on your own computer. It's built for depositions, meetings, interviews, sermons, podcasts, and any recording you need in writing — and it's free.
Transcribe audio and video files, or record live from your microphone — or capture a Zoom, Teams, or browser call — and transcribe as you go.
Sort out who said what and give speakers real names or roles like "Atty. Smith" or "Witness."
Tidy punctuation and grammar across the whole transcript automatically — without changing what was said.
From short overviews to detailed minutes with key points and action items. Long recordings are handled in sections automatically.
Search the transcript — including live, while you're still recording.
Save to Word (.docx), PDF, or subtitle/caption formats for video.
The app is free and works fully offline. It ships with a capable built-in engine. For even better speed and accuracy you can download optional upgrades or plug in your own AI account — but you never have to.
Click Live Record to capture and transcribe audio in real time — from a microphone, or from an app like Zoom, Teams, or a web browser. Choose "Record app / system audio" to capture the other participants, and leave "Also include my microphone" checked to capture your own voice too, so the whole conversation lands in one transcript.
For the cleanest capture of a specific app without affecting what you hear, you can install the free VB-CABLE tool — the app includes a built-in help button that walks you through it. This is optional; the standard "record app audio" option works without it.
After a recording, click Identify Speakers to auto-detect voices, correct any line, and give speakers real names. Pick a line and the app highlights other lines that sound like the same person, so you can fix several at once.
Telling voices apart from audio alone is hard, and no offline tool is perfect — especially with crosstalk or similar-sounding speakers. The app gives you a strong starting point plus easy manual controls to make it exactly right. Tip: if you know how many people spoke, choose that exact number — it gives far better results than letting it guess.
The app works well out of the box. For more speed or accuracy, there are a couple of optional add-ons.
A one-time ~35 MB download (still fully offline) that's noticeably more accurate at telling voices apart, especially with three or more speakers.
Have an API key for Claude, ChatGPT, or Gemini? Enable it for top-tier summaries and grammar, effortless handling of very long recordings, and smart speaker labels by role.
Cloud AI is completely optional and off by default. Nothing is sent anywhere unless you turn it on and enter your own key. Telling voices apart is always done locally.
Yes. After the one-time download on first launch, everything runs on your computer. The only exception is the optional Cloud AI feature, which you turn on yourself.
No — not unless you enable Cloud AI. By default, your recordings and transcripts never leave your computer.
Common audio and video formats including MP3, WAV, M4A, AAC, FLAC, OGG, MP4, MOV, M4V, MKV, and AVI.
Separating voices from audio is genuinely difficult. Use the exact speaker count when you know it, download the better speaker engine, and use the manual drop-downs and "find similar" to correct any lines.
The basic app uses your CPU. For very long jobs, enabling the optional Cloud AI feature (with your own API key) handles them faster and more easily.
There's a Donate button in the app if you'd like to chip in. It's appreciated but never required.
Free for Windows. The installer is signed and works fully offline after first launch. Mac version coming soon.