French audio to text
Drop a French recording in and get a French transcript, produced locally in your browser — nothing uploaded, no account, no per-minute charge.
French is among the stronger languages for Whisper, and clear recordings transcribe well including accents and elisions.
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: multilingual model, language set to French.
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 “French 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.
Liaison, elision and how they land in text
Spoken French joins words together in ways the written form does not — liaison across word boundaries, elided vowels, dropped ne in negation. The model writes standard orthography rather than a phonetic rendering, so j’sais pas generally comes out as written French, and the negative particle that was not pronounced is often restored.
That is usually what you want for a readable transcript. If you are doing linguistic work where the exact spoken form matters, an automatic transcript is the wrong starting point regardless of tool — it normalises by design.
Accents and punctuation
Accented characters are produced correctly, and French spacing conventions around punctuation are largely followed. Punctuation itself is predicted from the rhythm of the speech, so a speaker who runs sentences together will produce longer sentences in the transcript.
Proper nouns and institutional names are the usual corrections. They repeat, so fixing the first instance and searching for the rest is fast.
Translating into English
Switch the output to translate and the model produces English text directly from French speech, in a single pass. It only translates into English — French output from English speech is not something Whisper does.
The result is good enough to understand a recording and decide what matters in it. For anything being published or relied on, treat it as a rough gloss rather than a translation.
Things that save a re-run
- Leave detection on for a wholly French clip; set the language explicitly for short or mixed recordings.
- The model normalises to standard orthography — do not expect a phonetic record of what was said.
- Institutional names and acronyms are the most frequent corrections.
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 it translate French speech into English?
Yes — set the output to translate and it produces English text directly from the French audio in one pass. The reverse direction is not supported; Whisper only translates into English.
Are accented characters written properly?
Yes. The model produces standard French orthography with correct accents, predicted from the speech.
Does it transcribe informal spoken French accurately?
It normalises towards written French — elided forms are usually expanded and dropped negation is often restored. That makes a more readable transcript but a poorer record of the exact spoken form.
Is any of this processed on a server?
No. The multilingual model runs inside your browser and the audio never leaves your device.