FLAC to text
FLAC is lossless compression: the file is smaller than WAV but nothing has been discarded. Drop one in and it is transcribed on your device, with no upload.
It is the format of archives, field recorders and anyone who did not want to make a quality decision at capture time — which usually means the recording matters.
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 “FLAC to text”.
- 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.
Archive and research audio
People choose FLAC when the recording is meant to last: oral history projects, ethnographic fieldwork, band rehearsals, radio archives. Those recordings are often exactly the material that should not be uploaded — oral history in particular usually carries a consent form describing precisely who may hear it, and a cloud transcription service is rarely on that list.
Transcribing locally keeps the chain of custody intact. The file stays in your working folder, the transcript is written by your own machine, and there is no third-party copy to account for in a data-management plan.
Lossless does not mean easy
A perfect capture of a difficult room is still a difficult room. FLAC preserves whatever the microphone heard, including the reverberation, the air conditioning and the person talking over the interviewee. Recognition accuracy is set by those things, not by the codec — a lossless recording of a crowded café will transcribe worse than a modest MP3 of a quiet office.
Where lossless genuinely helps is repeated processing. If you are going to filter, denoise or normalise before transcribing, starting from FLAC means the cleanup is not compounding an encoder’s existing damage.
Large files and long sessions
FLAC files from field recorders are often high sample rate and multi-channel, which makes them large in memory once decoded. Everything is downmixed to 16 kHz mono here, but the decode happens first — so a 24-bit 96 kHz stereo hour is a lot of RAM before it is reduced.
For very long archive sessions, converting to 16 kHz mono WAV or FLAC first makes the whole thing lighter with no effect on the transcript. Splitting by tape side or session is usually a better structure for the output anyway.
Things that save a re-run
- Downsample long archive recordings to 16 kHz mono before transcribing — same accuracy, far less memory.
- Denoise before transcribing rather than after; the model cannot undo noise, and cleanup helps it.
- Keep the original FLAC untouched — export the transcript alongside it rather than re-encoding.
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 FLAC more accurate than MP3 for transcription?
Slightly, and rarely noticeably. The model consumes a 16 kHz mono representation that a good MP3 already preserves. FLAC’s real advantage is that it survives repeated editing and cleanup without degrading.
Do browsers support FLAC?
All current desktop browsers decode FLAC natively, so it works here directly. If an unusual variant fails, converting to WAV always works.
Why is my high-resolution file using so much memory?
Decoding happens at the file’s own resolution before it is reduced to 16 kHz mono. A 24-bit, 96 kHz stereo recording is roughly twelve times the data of the 16 kHz mono the model actually needs. Converting first avoids that peak.
Can I transcribe an archive without breaking a consent agreement?
That depends on your agreement, and this is not legal advice — but the technical position is clear: the file never leaves your machine, so no new party gains access to it. That is materially different from uploading it to a service, and it is usually the distinction such agreements turn on.