Transcribe a legal recording or client call
Client calls, witness interviews, recorded hearings and case-file audio transcribed on your own device, with no upload and no vendor holding a copy.
A working transcript for review and drafting — produced without routing privileged material through a third party’s servers.
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, timestamped 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 “Legal & client calls”.
- 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.
Privilege and third-party disclosure
Sending a recording of a privileged conversation to an outside transcription vendor is a decision with consequences that vary by jurisdiction and by the terms you have in place — which is precisely why firms have policies about it, and why those policies usually involve a signed agreement, a security review and a retention schedule. Getting a rough transcript quickly should not require any of that.
Here the recording is not disclosed to anyone, because it is not transmitted. The page loads a speech model into your browser and processes the file locally. There is no upload endpoint, no queue, no account and no retained copy. That does not answer your firm’s policy question for you, but it changes the facts the answer depends on.
Not a certified transcript
This produces a working draft, not a record. Court and deposition transcripts prepared by a certified reporter carry an attestation, a defined format and a person who can be examined about them; an automatic transcript carries none of that, and no jurisdiction accepts one in place of the certified version. Anything intended for filing, service or the record needs a human reporter.
Where a machine draft genuinely helps is everything before that: finding the three minutes that matter in a four-hour recording, drafting a chronology, preparing questions, checking whether a call contained what you remember it containing. Use the timestamps to cite back into the audio and verify before relying on any passage.
Recording law is not addressed here
Whether a conversation could lawfully be recorded in the first place is a separate question from how it is transcribed, and it varies sharply — consent requirements differ between US states and between countries, and the rules for a phone call, an in-person meeting and a hearing are not the same. This tool takes a file you already have and does not opine on how you got it.
If there is any doubt about the provenance of a recording, resolve that before transcribing it, not after. A transcript makes audio much easier to circulate, which is a benefit when the recording is sound and a problem when it is not.
Things that save a re-run
- Use the timestamped export so every passage can be cited back to a position in the audio.
- Names and case citations are the most frequent errors — check those specifically before circulating.
- For long hearings, transcribe once and search the transcript rather than scrubbing the recording.
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
Can this be used for a deposition or court transcript?
No. Certified transcripts require a qualified human reporter and carry an attestation that automatic output cannot. This is for working drafts — review, chronologies, finding the relevant passage — not for anything that goes on the record.
Does the audio go to a server?
No. The site is static files with no upload endpoint. The speech model runs in your browser and the file stays in the tab. You can verify it by transcribing with the network disconnected once the model is cached.
Is using this consistent with my firm’s confidentiality obligations?
That is a question for your firm — this page does not give legal advice. The relevant technical fact is that no third party receives the recording, because it is never transmitted. Whether that satisfies your particular obligations is for you and your risk or compliance function to determine.
How well does it handle poor-quality call recordings?
Telephone audio is narrowband and often compressed, which costs accuracy — expect a usable but imperfect draft, worse where people speak over each other. It is still far quicker than listening to the whole recording, provided you verify anything you intend to rely on.