Transcribe a therapy session without it leaving your device
Session recordings are among the most sensitive audio anyone makes. This tool transcribes them inside your browser — the file is never uploaded, and there is no account holding a copy afterwards.
Useful for supervision notes, training review and process notes where a written record helps and a cloud transcription service is not an option.
Language and translation need the Multilingual model.
Drop an audio or video file
or , paste with Ctrl + V, or
- MP3
- MP4
- WAV
- M4A
- MOV
- WEBM
- OGG
- FLAC
- AAC
The file is read by this page and never uploaded — no account, no size limit but your own RAM.
Preset: balanced English model, plain-text output.
Download
Click a line to jump the audio there. Click the text to fix a word — your edits go into every download.
How it works
- Open this page — the tool is already set up for “Therapy sessions”.
- Drop the file in, browse for it, or paste it with Ctrl+V.
- The model runs on your own device; the transcript appears as it goes.
- Correct anything misheard, then download text, subtitles or notes.
Why the upload step is the problem
A therapy recording is a detailed account of someone’s mental health, usually named, often including third parties they have talked about. Handing it to a transcription service is a disclosure of that material to an organisation the client has never heard of, governed by a contract they have not seen, retained for a period they were not told about. Whether any particular service is trustworthy is beside the point — the disclosure itself is the event.
Nothing is disclosed here because nothing is transmitted. The speech model is downloaded to your browser once and runs on your own processor; the audio is read from disk by the page and stays in the tab’s memory. Closing the tab discards it. You can verify this by transcribing with your network disconnected, which works once the model is cached.
What this does not do
It does not make a recording compliant, and it is not compliance advice. Whether you may record a session at all, what the client must be told, how long a recording may be kept and where it may be stored are questions for your professional body, your supervisor, your employer’s policies and — in regulated settings — your data protection officer or counsel. Those obligations exist regardless of which software touches the file.
What it does change is one specific risk: the transcription step no longer requires giving the recording to anyone else. That is a narrow claim, and it is the only one being made.
Practicalities for session audio
Two people in a quiet consulting room is close to the best case for speech recognition, and the Balanced model handles it well. The recurring errors will be clinical terms, medication names and any name that is not common — fix the first occurrence and use the search box to find the rest.
The manual speaker labels are worth using here: click the first client turn, label it, then label your own first turn, and the alternation is captured. For supervision notes, the timestamped text export lets you point a supervisor to a moment without sending the audio itself.
Things that save a re-run
- Transcribe with the network disconnected if you want a demonstrable record that nothing could have been sent.
- Delete the source recording on the schedule your policy requires — the transcript does not remove that obligation.
- Medication and diagnostic terms are the most common errors; check them specifically.
Read the transcript against the audio before you rely on it. Every speech model — this one and the paid cloud ones — mishears names, numbers and crosstalk, and it does so confidently. The player and editor here exist so that check takes minutes rather than an afternoon.
Frequently asked questions
Is this HIPAA or GDPR compliant?
Compliance is a property of your whole process, not of one tool, and this page cannot tell you whether yours qualifies — ask your DPO, compliance lead or counsel. The technical fact that is relevant to that assessment: the audio is not transmitted to any server, so there is no processor to contract with and no third-party copy to account for. Everything else — consent, storage, retention, breach procedures — is unchanged and still yours.
Where is the recording stored while it is being transcribed?
In the browser tab’s memory only. It is not written to a server, and it is not saved into the site’s cache. Closing the tab discards it. The only thing cached is the speech model itself, which contains none of your data.
Can anyone at TranscribeLocally see the audio or the transcript?
No. There is no mechanism by which we could — the site is static files, with no upload endpoint and no server-side processing. We cannot see what you transcribe any more than a text editor’s publisher can see what you type into it.
Is it accurate enough for clinical notes?
It is accurate enough to be a first draft that you then read against the audio. It will mishear medication names, diagnoses and unusual proper nouns. No automatic transcription — local or cloud — should go into a clinical record unread.