Private transcription · Your server only

Your voice. Your server. Period.

Whisper-grade accuracy for people whose audio cannot leak. Processed on a single server we own. Deleted the moment we're done. No OpenAI. No Google. No third-party clouds.

Try free → How it works
● 99 languages · Whisper large-v3 ● No file size cap on paid tiers ● Built for lawyers, therapists, doctors, journalists
🎙
Drop an audio file, or click to choose
MP3 · M4A · WAV · OGG · FLAC · WebM · MP4 (audio track) — up to 200 MB
Sign in to start
How it works

Three steps. No detours.

Most transcription services route your audio through OpenAI, Google, or Azure. We don't. Here's exactly what happens.

1.

You upload

Your file travels straight from your browser to our single server over TLS. We never hand it to a third party.

2.

We transcribe locally

OpenAI Whisper large-v3 runs on hardware we own. The model is on-disk; nothing is sent to any AI API.

3.

We delete

The moment your transcript is returned, the audio and the transcript are deleted from disk. We keep usage counts, not content.

Built for

People whose audio can't leak

You already know cloud transcription isn't an option. Here's a tool that respects that.

Lawyers

Client calls. Depositions.

Privilege survives only if the audio doesn't leave a server you trust. Run drafts here, edit elsewhere.

Therapists · Counselors

Session notes.

SOAP notes from your recorder. No PHI in OpenAI's training set. Yes, we'll sign your DPA.

Doctors · Clinicians

Dictation.

Chart faster without trusting Big Tech with patient voices. A single server, single key, single tenant on paid plans.

Journalists

Source interviews.

Protect your source the same way you protect your notes. Their voice doesn't go to ChatGPT.

Pricing

Honest tiers. No surprise fees.

You can try it free, no card. Upgrade when it becomes part of your workflow.

Free
$0

For trying it out today.

  • 3 transcriptions per day
  • Up to 25 minutes per file
  • All 99 languages
  • .txt and .srt download
Try free
Solo
$7 / month

For one practitioner. Cancel anytime.

  • 10 hours of audio per month
  • Files up to 2 hours each
  • Speaker labels (coming soon)
  • Priority queue
  • Email support
Get notified at launch
Lifetime
$69$129

Pay once. Use forever. First 50 only.

  • Everything in Solo, forever
  • All future features included
  • Locked-in privacy promise
  • Founding member badge
Reserve a seat

Team plans (3-10 seats, single-tenant server) — contact us.

FAQ

Straight answers.

Where exactly is "our server"?
A dedicated machine we own, located in South Korea, fronted by Cloudflare TLS termination. Your file travels in over HTTPS, goes through Whisper large-v3 in-process, and the file plus transcript are removed from disk before the response is sent. We log timestamps and file sizes for billing, not content.
What model does the transcription?
OpenAI Whisper large-v3, run via the open-source faster-whisper engine. The model weights live on our disk; we never call the OpenAI API.
How does this compare to MacWhisper (on-device)?
MacWhisper is excellent for single-user Mac workflows: the model runs on your laptop and your audio never leaves. Silent Scribe is the alternative when you want the same privacy posture but need a hosted endpoint your team can hit from any device — phone, browser, automation. We're explicit about the trade-off: your audio still travels to a server, just one we control end-to-end.
Are you HIPAA / GDPR compliant?
We are not currently advertising HIPAA Business Associate or GDPR Article 28 contracts — formal compliance is a paperwork milestone we'll reach when there's clear demand. What we can offer today is a documented data flow, immediate-deletion policy, and a DPA on request for Team plans. If you need a signed BAA today, MacWhisper on-device or an enterprise vendor is the safer fit.
How many people can use it at once?
Free and Solo plans share a single processing queue. Most files take 8-15 seconds per minute of audio. Team plans get a dedicated single-tenant machine on request.
Can I run this on my own hardware?
Yes — the backend stack (FastAPI + faster-whisper + Cloudflare Worker gateway) is open at github.com/vis555540-ops/ledeuxions-site. Self-hosted licensing is available for organizations on request.