Frequently asked questions
Answers you can check rather than believe: cut your network connection, open your browser's “Network” tab, read the page's content security policy.
- Is my recording sent anywhere?
- No. Your browser reads the file and transcribes it on your computer. The content security policy forbids the page any connection to another domain, and this site's server has no address that could receive a file. The Privacy page explains how to check.
- So what is downloaded?
- The pages, the code and, the first time, the compute engine and the chosen model, all from this site. The model is then stored on your computer and does not download again.
- Does it work without a connection?
- Yes, once the model is loaded. You can disconnect before dropping your file, and even reload the page: it comes back from your browser's cache.
- What computer do I need?
- A recent computer with 8 GB of memory and Chrome or Edge gives good results. A graphics card the browser can use (WebGPU) makes the Balanced and Accurate models much faster. Chrome and Edge are the recommended browsers; on Firefox and Safari, it is not guaranteed to work.
- What about a phone?
- The Fast model runs there, slowly. Klarword is designed for a workstation: heavy models, intensive computation, proofreading at the keyboard. On mobile, the site tells you so and offers to open the address on your computer.
- Which formats can I import?
- MP3, M4A, WAV and MP4, and most formats your browser can play (OGG, WebM, FLAC, MOV depending on the system). The sound track of a video is extracted on the spot.
- Which languages are recognised?
- English, French and about fifteen other languages, with automatic detection. Quality is better with the Accurate model, especially outside English.
- How long does a transcription take?
- It depends on the model and the computer. On a recent laptop, an hour of recording takes about twenty minutes with the graphics card; the measured figures are on the home page. While it works, the tool shows the actual speed and the estimated time left.
- Is the transcript accurate?
- It is good on a clean recording, and fallible on proper names, overlapping voices and noisy rooms. Proofread it before sharing: that is what the editor is for.
- Does Klarword recognise who is speaking?
- Yes, on your computer: once transcription has finished, “Identify speakers” finds the voices (automatically or for a given number), then you rename each one just once. Two small models totalling 33 MB do the work, on the processor. Exports use the names: paragraphs per person in Word, voice tags in WebVTT. Check the result: two similar voices can be mixed up.
- Can Klarword write minutes?
- Yes, on computers with a recent graphics card: a small language model (Qwen3.5 2B, 1.4 GB downloaded once from this site) writes a summary, key points, decisions and actions, on your computer. The text goes nowhere. It is a draft: a model this size may leave out a point or mix up who said what, so check it against the transcript.
- Can I try it without downloading anything?
- Yes: the tool offers an example, one minute of an interview already transcribed. You can listen to it, correct it, export it or burn subtitles into it without downloading a model; only the audio (430 KB) and its text come from this site.
- Can Klarword translate subtitles?
- Yes, on computers with a recent graphics card, with the same small language model as the minutes. The lines are translated on your computer, at the original timings, into English, German, Spanish, Italian, Dutch, Portuguese or French. It is a machine translation: proofread it, especially proper names.
- How do I anonymise a transcript?
- The “Pseudonymise” panel finds likely proper names and suggests them; the ones you tick are replaced by “Person A”, “Person B”, in the text and in the speaker names alike. Detection follows simple rules: proofread, and fill any gaps with “Find and replace”.
- Can I transcribe several files at once?
- Yes: drop several, and Klarword transcribes them one after another with the chosen model, then gives you a ZIP file with the formats you ticked (Word, SRT, WebVTT, text, JSON). Each transcript can then be reopened in the editor for proofreading.
- What can burned-in subtitles look like?
- Four styles: white outline, dark banner, large text for social media, and high legibility (Atkinson Hyperlegible font, yellow on black). Size, font, position (bottom, top, or automatically above text already in the picture), one colour per voice, and a vertical 9:16 crop. The preview shows the frame exactly as it will be encoded.
- Can I make bilingual subtitles, or clips for social media?
- Yes. A translation made on this computer can stay next to the original: subtitles then show both languages, the translation in italics under the original, in SRT, WebVTT, TTML, SBV and ASS as well as in the burned-in video. And each chapter, or a passage you choose, becomes a short subtitled video, with its title at the top of the picture for the first seconds, in the video's format or vertical 9:16.
- Can I blur faces, or mask names in the audio?
- Yes, on this computer. When burning in, faces can be turned into a mosaic: a small model (232 KB) finds them on every frame, and the vertical crop can follow the face of whoever is speaking. After pseudonymisation, Klarword offers, without imposing it, a copy of the audio in which each name is covered by a beep or silence. In both cases, watch or listen again before publishing: a face in profile or a misheard name can slip through.
- Can I protect a piece of work, or see what was corrected?
- A .klarword file can be protected by a password (AES-256, the key derived from the password with PBKDF2): without it the file does not open, and it cannot be recovered. The “Compare two versions” panel shows what changed between the model's transcription and your corrections, or against another file, and “Speaking time” counts who speaks how much.
- How do I quote a precise passage?
- Each word carries the moment it was spoken: Ctrl+click on a word to hear it again. The “excerpt” button on a line saves the passage as WAV, with half a second of margin, and tags (“to quote”) find the marked passages, which can be exported as a CSV table. Quotes suggested by the model are shown only if they appear word for word in the transcript.
- Does Klarword check subtitles?
- Yes, before export: it flags subtitles that read too fast (more than 17 characters per second), those that stay on screen too briefly or too long, and lines that are too wide, and each flag takes you to the passage. These are the usual platform standards; you can still export, knowingly.
- Can I pass a piece of work to a colleague?
- Yes, without a server: “Save the project” produces a .klarword file containing the recording, the transcript, the speakers, the annotations and the minutes. It reopens in Klarword on another computer, as is. You send it like any other file in your organisation.
- Can Klarword be installed?
- Yes, with Chrome or Edge: the tool's “Install Klarword” button opens it in a separate window, and your audio and video files then open with a double-click. It is the same site, working the same way: nothing else is installed, and nothing more is sent.
- Is my work saved?
- Not without your agreement: by default, nothing is kept when the tab closes, and the browser warns you if you leave without saving. If you tick “Keep this work on this computer”, the recording and its transcript are stored on your computer, never sent, and you find them at the start of the tool. A JSON, WebVTT or SRT export also reopens in Klarword, with the recording.
- Where are the models stored, and how do I delete them?
- In the private space the browser keeps for this site. The tool's “On this computer” panel lists and deletes them; clearing the site's data in your browser settings erases them too.
- Who must obtain the consent of the people recorded?
- You do. Klarword is a tool; informing people and obtaining their agreement remains the responsibility of whoever records. See the terms of use.
- Can we install it on our internal network?
- Yes: that is self-hosting. The whole site, models included, sits on your intranet and works with no outside connection at all.
Still a question?
What each feature does not do is written next to it, in the tool, and the Privacy page sets out the checks, the counter and what we do not guarantee. Otherwise, write to us: contact@klarword.com.