How to transcribe an audio file to text
Whether it's a meeting you recorded, a lecture, an interview or a memo on your phone, the steps to transcribe an audio file are the same: get the file, upload it, check the text. This page does the middle part for free, with a timestamp on every line.
The rest of this guide covers everything around it: which format to use, how to record so the transcript comes out clean, and what to do with the text once you have it.
Audio formats you can upload
- MP3: the most common audio format, used for podcasts, downloads and most exports. Format tips on the MP3 to text page.
- M4A: what iPhones, Macs and many recorder apps produce. See M4A to text, or the voice memo guide if your recording is in the Voice Memos app.
- WAV: uncompressed audio from dedicated recorders, audio interfaces and editing software.
- OGG and Opus: common on Android recorders, and the format of most WhatsApp voice notes.
- WEBM: what many browser-based recorders save.
Video files go to MP4 to text. If your audio is in a format that isn't listed, such as AIFF, FLAC or WMA, export it as MP3 or M4A from your recorder or editing app first.
Choosing a format before you upload
If you get to decide how the file is saved, pick a compressed format. Speech recognition doesn't need studio-quality sound, and smaller files upload faster and fit within the free tool's 4 MB limit. A rough guide to how much audio fits in 4 MB:
- WAV at CD quality (stereo) uses about 10 MB per minute, so 4 MB is under half a minute. Convert long WAVs to MP3 first.
- MP3 at 128 kbps uses about 1 MB per minute, so roughly four minutes.
- MP3 or M4A at 64 kbps, which is plenty for a voice, fits about eight minutes.
Whatever the format, the free tool transcribes the first 10 minutes of what you upload.
How to transcribe audio to text with this tool
- Drop your audio file into the box at the top of this page, or click the box to choose it.
- Press Transcribe. A speech recognition model listens to the recording and detects the language by itself, whether it's English, Spanish, French, German, Portuguese, Italian or another major language.
- After a few seconds, you'll see the transcript split into lines, each with the time it starts.
- Copy what you need, or download it as .txt, .txt with timestamps or .srt with a free Claras account.
Meetings, lectures and interviews
Meetings. Recordings from Zoom, Teams or a phone in the middle of the table all work, as long as you can save the audio as a file. Write up decisions and owners from the transcript while the meeting is still fresh, and keep the timestamps so anyone can check the exact words.
Lectures. Sit close to the speaker, or ask if you can leave your phone near the lectern. A clear lecture transcript turns into study notes quickly, and searching it for a term the professor mentioned beats scrubbing through an hour of audio.
Interviews. The transcript comes as timed lines without speaker names, so add initials as you edit. Before quoting anyone, use the timestamp to listen to the line once more.
Podcasts and voice notes. A podcast transcript is the raw material for show notes and quotes, and a transcribed voice note means you can read a long message instead of listening to it in a quiet office. Opus voice notes from chat apps upload just like any other file.
Recording tips for a cleaner transcript
- Record in a quiet room. Fans, traffic and coffee machines cost more accuracy than any format choice.
- Keep the microphone close to whoever is talking. Distance adds echo, and echo blurs words.
- Ask people to take turns. Overlapping voices are the hardest thing for a speech model to untangle.
- Make a ten-second test recording and play it back before an important session.
- Plan to check proper names, numbers and jargon. Those are the words most often misheard.
Checking and editing the transcript
Even a good transcript deserves a quick review before it goes anywhere important. A practical routine:
- Read it through once at normal speed and mark anything that looks odd.
- For each marked line, use the timestamp to jump to that moment in your audio player and listen.
- Fix names, numbers and terms, and add speaker initials if there's more than one voice.
- Remove filler words if the text will be published, but keep direct quotes exact.
For notes only you will read, you can usually skip straight to the third step.
Your audio and privacy
Recordings often hold things you wouldn't post publicly, so it's fair to ask what happens to them. A file you upload here is sent once for transcription and isn't kept afterward. The providers that process it keep no copy, and your audio is never used to train AI models.
If a recording includes other people, it's good practice to tell them you're recording and that you'll turn it into text.
What to do with the transcript next
A transcript is a starting point. Here's what people commonly do with it in Claras:
- Summarize it. Every transcript can get an AI summary with a TL;DR and key points, so an hour of audio becomes something you can read over coffee.
- Ask it questions. The Ask AI chat answers questions like "what did we decide about the budget?" and cites the timestamp, so you can check the source.
- Export it. Download TXT for documents, PDF for sharing, or .srt for subtitles.
- Send it to another AI. Pass the transcript to ChatGPT, Claude, Gemini or NotebookLM to draft an email, a blog post or flashcards from it.
The same works for online videos: Claras transcribes YouTube, Instagram and TikTok from a link, and the subtitle downloader grabs existing YouTube captions.
Free tool, free account and paid plans
Your first file here is free with no account, up to about 4 MB and the first 10 minutes. A free Claras account adds downloads and 1 hour of audio and video file transcription a month, with no card. Files several hours long work in the Claras web app at app.meetclaras.com.
If you transcribe regularly, Pro is $5 a month for 10 hours of file transcription, and Unlimited is $12 a month for 100 hours. Both include a 14-day free trial. The pricing page has the full comparison.