Privacy

What is uploaded

Nothing leaves your browser until you press Upload & transcribe. Choosing a file, playing it back, and reading this page happen locally.

When you start a transcription, the recording is sent over HTTPS to our server and forwarded to AssemblyAI, our transcription provider, which processes it in the United States by default. We use the provider's flagship speech model with speaker diarization, falling back to Universal-2 for languages the flagship model does not cover, so the transcript includes speaker labels and timestamps.

The one-time agreement you accept when you sign in covers this processing; the terms are on the Terms page.

Do not upload recordings you do not have permission to share. The provider API key is held only on our server and is never exposed to the browser.

Retention and third-party deletion

AssemblyAI's asynchronous pre-recorded API does not offer zero data retention. By default, uploaded audio is deleted within 24–48 hours and transcribed results after 30 days. We request deletion of the vendor-side transcript as soon as the completed result has been saved, so it is normally removed immediately rather than after the default window.

Account-level controls on the provider side — a time-to-live as low as one hour and opting out of model training — must be enabled in the provider's dashboard. Opting out of model training requires a paid provider account. Vendor deletion is subject to the provider's own processing times.

What we store

We never store the audio file. In our database we keep job metadata (status, file name, file size, provider job ID, the terms version and time you accepted, audio duration and credits charged), the resulting transcript segments, and a daily per-account counter used to enforce quota limits. The transcript segments are tied to your account so you can retrieve a long-running job after reloading the page.

Canceling a job or deleting its saved copy removes the stored transcript and requests deletion of the provider-side transcript. You can also download your transcript and stop using the service at any time.

Accounts, credits and quotas

Choosing and uploading a recording requires no account. Every transcription API call — submit, status poll, and deletion — requires a signed-in session, which our server verifies before reading any audio bytes.

Transcription spends credits: 1 credit covers 30 minutes of audio, billed in 30-minute blocks. New accounts receive 2 free credits once. Daily per-account limits on file count and total megabytes also apply, because cloud transcription is billed per audio hour.