Transcribe a recording and get a summary you can use
To transcribe a recording in Claras, upload the audio or video file. You get a timestamped transcript, a summary with chapters, and a chat that answers from what was said, for files up to 6 hours long.
The recording you meant to write up
A meeting, an interview, a talk you gave, a voice note to yourself. Recording it was easy. Turning it into text means playing a few seconds, typing, rewinding and playing again, and that takes far longer than the recording itself.
So most recordings sit in a folder. When you need one line from them, you can't search audio, and you end up scrubbing through an hour to find two minutes.
Who this is for
Anyone with their own audio or video files: meeting and class recordings, interviews for an article or a project, podcasts in progress, voice memos, webinars you saved. If your recordings are calls with leads or customers, see how sales teams use it, since that page covers their questions.
Claras works from a file you already have. It doesn't join meetings, record them or transcribe while you talk.
How to transcribe a recording in Claras
You need a free account and the file. Video files can be MP4, MOV, WebM or MKV. Audio files can be MP3, M4A, WAV, FLAC, AAC, OGG or Opus.
- Sign in, then open the Video or Audio tab and drop in your file.
- Pick the language for your summary. For a file of 30 minutes or more, tick "Email me when it's done" if you like.
- Press Analyze with AI and keep the tab open while it transcribes.
- Read the summary, search the transcript, then copy or download what you need.
A worked example: a 45-minute meeting recording
Say you saved a 45-minute team meeting from Zoom to your computer and need the decisions written down. Here is what happens after you drop the file in.
1. Claras checks the file
It reads how long the recording is from the file's headers, then checks your transcription hours. 45 minutes fits in the free plan's 1 hour. If a file is longer than the hours you have left, Claras says so before it starts, with the hours remaining.
2. Your browser reads the audio
The video part is skipped and only the audio is read, a piece at a time, so a large video file is never loaded or uploaded in full. The audio is cut into slices of a few minutes, each overlapping the next by a few seconds so a sentence at a cut is heard whole. WAV and FLAC are shrunk first, in your browser.
3. The slices are transcribed
The first slice goes alone, so the AI can work out the language. The rest go four at a time, and the button shows the progress as a percentage. Finished slices are kept in your browser, so if the tab closes, choosing the same file again picks up where it stopped.
4. The summary is written
Once the transcript is stitched together, Claras writes chapters, a short opening, the key takeaways and a summary of each section, in the language you picked. Click a chapter and your recording plays from that moment, straight from your own disk.
5. You get what you came for
Ask AI questions like "What did we decide about the launch date?" or "What were the next steps?". The answer comes from the transcript, usually with the time it was said. "Who agreed to send the report?" only works if the name was said aloud, because the transcript has no speaker labels. Then copy the summary into your notes or send it to the team.
What you get from one recording
- The transcript. Every spoken line with its timestamp. Search it, click a line to play from there, and copy it with or without timestamps.
- Downloads. The transcript as a text file, with or without timestamps, the summary as a text file, and .srt subtitles you can load into a video editor.
- A summary. A short opening, 4 to 8 key takeaways and a section for each part of the recording. It's built from the whole transcript, read in 20-minute stretches.
- Chapters. Titled chapters with a one-line description each, for recordings of 3 minutes or more. Copy them as an outline.
- Ask AI. Questions answered from the transcript, plus three suggested questions.
- Send to AI. Copy the transcript and open Claude, ChatGPT, Gemini, Gemini Notebook or Perplexity, for anything Claras doesn't do itself.
Private by default
An uploaded recording gets no public share link, unlike a YouTube summary. It's saved to your History, where you can reopen the transcript and summary later. The file isn't kept, so to play it again, choose it once more.
Tips for better transcripts
Free tool or the full app?
For a short clip where you only need the words, the free tools are quicker: MP3 to text, M4A to text, any audio file, voice memos and short MP4 clips. They need no account, but each takes one file up to 4 MB and shows the transcript of its first 10 minutes. Copying or downloading the result needs a free account, and after one file the tool asks you to sign in.
Use the app for anything longer, or when you want a summary, chapters, Ask AI and a saved copy.
| Free tools | Claras app | |
|---|---|---|
| Account | None for the first file | Free account, no card |
| File limit | 4 MB | Up to 6 hours, no size cap |
| Transcript shown | The first 10 minutes | The whole recording |
| Summary, chapters, Ask AI | No | Yes |
| Uses your hours | No | Yes |
Trim the dead air
Hours are counted by the length of the audio. If a recording starts with ten minutes of waiting for people to join, cutting it saves those minutes.
Say names and decisions out loud
With no speaker labels, "Maria will send the report by Friday" is far more useful in a transcript than "I'll do it".
Keep one language per recording
The language is set from the first few minutes. A recording that starts in English and switches to Spanish may come out poorly after the switch.
Ask for the format you need
Custom summary instructions in Settings, up to 500 characters, shape every summary after you set them. For meetings, try "List the decisions and action items in each part".
If the format isn't read
If Claras says your browser can't read the file's audio, export or convert it to MP3 and try again. Name the file well too: the name becomes its title in your History.
Limits to know about
Length and hours
- Up to 6 hours per file. The app sets no file size limit.
- Uploads use transcription hours: 1 on the free plan, 10 on Pro and 100 on Unlimited, over a rolling 30 days. Each file also counts as one video, and the free plan has 3 a month.
- The pricing page has the details of each plan.
Keep the tab open
Transcription runs in your browser tab. You can use other tabs, but closing this one stops the job. Choose the same file again to continue.
What the transcript leaves out
- No speaker names. Lines are timestamped but not labelled by person.
- Quiet or overlapping speech, strong background noise and people far from the microphone are often missed or misheard.
- Names, numbers and technical terms can come out wrong. Check anything that matters against the recording.
- The summary is the AI's reading of the transcript, so it can misstate a point or leave out a condition.
What Claras doesn't do
- It doesn't join or record meetings, and there's no Zoom, Teams or Meet connection. Save the recording, then upload it.
- No live transcription while you speak.
- The Chrome extension works on YouTube only. Uploads go through the web app.
Privacy and consent
The file isn't stored. Its audio is sent in slices to an outside AI transcription service set up not to keep it or use it for training. Please only upload recordings you have the right to use, and ask the people in them first.
Related
Compare plans to see the transcription hours on each.
Questions about transcribing recordings
How do I transcribe a Zoom recording?
Save the recording as a file first, then upload it in the Video or Audio tab. Claras doesn't connect to Zoom or any other meeting tool, so it works from the file you download. A video file or an audio-only file both work, since Claras only reads the audio.
Is there a file size limit?
The app doesn't set one. Your browser reads the audio out of the file a piece at a time, so the limit is length: up to 6 hours per file, and no more than the transcription hours you have left. The free tools are different: they take one file up to 4 MB.
Can I transcribe a recording for free?
Yes, two ways. A free account includes 1 hour of transcription every 30 days and 3 videos a month, with no card. Or, for a short file, the free audio file transcriber needs no account: it takes a file up to 4 MB and shows the transcript of the first 10 minutes.
Does the transcript say who said what?
No. Claras gives you timestamped lines of what was said, without speaker names. If you need to know who said something, listen at that time, or make sure people say names and decisions out loud.
Which languages can it transcribe?
The AI works out the spoken language from the first few minutes of the file, so you don't need to set it. The summary, chapters and Ask AI answers come back in the language you pick, one of 15. A recording that switches between languages may come out less well.
Is my recording stored?
No. The file stays on your computer. Its audio is sent in slices to an outside AI transcription service set up not to keep it. The transcript and summary are saved to your History so you can find them again.
What happens if the upload stops halfway?
Choose the same file again in the same browser. The slices already transcribed are kept there, so Claras carries on from where it stopped instead of charging those minutes twice. A private window may not keep them.
Can I get subtitles from my recording?
Yes. In the web app, download the transcript as an .srt file, which most video editors and players can load as subtitles. You can also download plain text, with or without timestamps.
Watch less. Learn more.
Paste a video link and get the summary, chapters and answers in seconds.