MP3 to transcript: how to convert MP3 to text
MP3 is the format most audio ends up in: podcast episodes, recorded interviews, lectures you downloaded, the audio track you pulled out of a video. The words are all in there, but you can't search, skim or quote them until they're written down. This page turns an MP3 into a transcript with a timestamp on every line.
Below you'll find how the converter works, what fits in the free version, and a few habits that make the difference between a transcript you can use as is and one you have to fix line by line.
How to convert an MP3 to text here
- Find the MP3 on your computer or phone. On a phone, it's usually in your Files or Downloads app.
- Drop it 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 file, writes down every word and notes when each line starts.
- Read the transcript right on the page. Copy it, or download it as .txt, .txt with timestamps or .srt with a free Claras account.
You don't need to pick a language. The model detects it by itself and handles English, Spanish, French, German, Portuguese, Italian and most other major languages.
What fits in the free MP3 to text converter
Your first file is free with no account. The tool takes MP3 files up to about 4 MB and transcribes the first 10 minutes. If the recording runs longer, you'll see the first 10 minutes and a note telling you how long the full file is.
How much audio fits in 4 MB depends on the bitrate the MP3 was saved at. A 128 kbps MP3, a common default, uses close to 1 MB per minute, so 4 MB is about four minutes. A spoken-word MP3 saved at 64 kbps uses half that, so roughly eight minutes fit. Speech doesn't need a high bitrate, so if you control the export, 64 kbps mono is a sensible setting for anything you plan to transcribe.
Transcript or subtitles: pick the output you need
Converting an MP3 to text can mean two different things, and the download options cover both.
- Plain .txt is the words as one continuous text. Use it when you want to read, edit or paste the content into a document.
- .txt with timestamps keeps the time each line starts. It's the best choice for interviews and research, because you can jump back to the exact moment in the audio to check a quote.
- .srt is a subtitle file. Each line has a start and end time, so video editors and players can show it as captions. Use it if the MP3 is the soundtrack of a video you're publishing.
Podcasts and interviews
Podcasters use an MP3 transcript to write show notes, pull quotes for social posts and give listeners a text version of the episode. With timestamps, you can scan the text for the strongest five minutes instead of listening to the whole hour again.
For interviews, the transcript comes as timed lines rather than a script with names, so speakers aren't labeled. A quick pass to add "Q:" and "A:" or each person's initials is usually enough. If you're quoting someone in an article, use the timestamp to listen to the line once more before you publish.
Other recordings that often arrive as MP3
- Lectures and courses. Downloaded course audio and recorded classes are often MP3. Search the transcript for the term you need instead of scrubbing through the file.
- Dictation. Many dictation recorders and apps export MP3. The transcript gives you a first draft of the letter, report or chapter you talked through.
- Sermons and talks. Churches, clubs and conferences often share recordings as MP3. A transcript makes the talk easy to quote or to share as a written recap.
- Audio pulled from a video. If you saved a video's soundtrack as an MP3, the transcript works the same, and the .srt file can go right back onto the video.
Music, intros and silence
This converter is built for speech. Songs, lyrics and talk over a loud music bed are hard for any speech recognition model, and the results can be patchy. Stretches with no speech at all are simply left out of the transcript.
If your MP3 opens with a long music intro, trimming it before you upload helps twice: the file gets smaller, and more of the free 10 minutes goes to the part where people are talking.
Tips for a more accurate MP3 transcript
- One speaker at a time. Crosstalk is where transcripts get messy. When two people talk over each other, the model may only catch one of them.
- Get the mic close. A phone on the table a foot away beats a laptop mic across the room.
- Keep background noise down. Cafés, fans and street noise cost more accuracy than the file format ever will.
- Avoid re-compressing. Every time an MP3 is converted into another MP3, it loses a little detail. Export from the original recording if you still have it.
- Check names and numbers. Proper names, brands, figures and technical terms are the words most likely to be misheard. Give them a quick look before you share the text.
If your file isn't an MP3
This page only accepts MP3 files. For recordings from an iPhone or Mac, use M4A to text, and if you're starting from the Voice Memos app, the voice memo guide shows how to get the file off your phone. Video files go to MP4 to text, and WAV, OGG or WEBM files work in the general audio file transcriber.
If the audio is already online, you may not need a file at all. Claras can transcribe YouTube videos, Instagram reels and TikTok videos straight from the link.
Full episodes and longer recordings
The free tool is meant for short clips. For a full podcast episode, a lecture or a long interview, the Claras app transcribes files several hours long. Along with the timestamped transcript, you get an AI summary with a TL;DR and key points, plus an Ask AI chat that answers questions about the recording and cites the timestamp it found the answer at.
From there you can export to TXT, PDF or .srt, or send the transcript to ChatGPT, Claude, Gemini or NotebookLM. A free account includes 1 hour of audio and video file transcription a month, with no card. If you transcribe every week, Pro is $5 a month for 10 hours and Unlimited is $12 a month for 100 hours. Both come with a 14-day free trial, and the details are on the pricing page.