Audio to Text Converter
Upload an interview, a lecture, a meeting or a voice note and get back a written transcript. Speech recognition writes out what was said, with punctuation, optional timestamps and speaker labels, as a document or as subtitles for video.
Audio to Text
or drop MP3, WAV, M4A … files here ·
Max 95 MB per file· Up to 20 files· Deleted after 30 minutes
Files are processed on our servers and deleted after 30 minutes.
Subtitle lines longer than this are split. 42 is the common broadcast standard.
Optional. Seconds (e.g. 75) or minutes:seconds (1:15). Leave empty to transcribe from the beginning.
Optional. Same format as the start time. Leave empty to transcribe to the end.
How to use this tool
- 1
Upload audio or video
Add MP3, M4A, WAV, AAC, FLAC, OGG, Opus, AMR or other audio files, or MP4, MOV, MKV, WebM and AVI videos. Up to 20 files, 95 MB each.
- 2
Choose language and output
Spoken language starts on Auto-detect. Pick Output: plain text, Word, PDF, SRT or VTT subtitles, or JSON with word timings.
- 3
Transcribe and download
Press Transcribe; the button also names the output you chose. Open the transcript, proofread names and numbers, and download it.
Audio to text conversion, also called transcription, turns recorded speech into written words. Upload the file, keep Spoken language on Auto-detect or pick the language, choose an output, and press Transcribe.
Transcribe audio to text: how speech recognition works
Speech recognition listens to the recording in short windows, turns each one into a picture of its frequencies, and lets a trained model decide which words most likely produced that sound. Modern models learn from very large amounts of transcribed speech, so they handle different voices and accents far better than the dictation software of a decade ago. Punctuation and capital letters are predicted from the flow of the sentence; turn Punctuation off if you want the raw words only.
The language setting matters. Auto-detect listens to the first 30 seconds and then uses one language for the whole file. If the recording starts with music, a jingle or silence, or switches between languages, choose the Spoken language yourself. The list covers 33 languages, from English, Spanish and Arabic to Hindi, Tamil, Japanese and Ukrainian.
People reach this tool from many directions: a journalist with an interview on a phone, a student with a recorded lecture, a team with a meeting export, a creator who needs captions. The two most common file types have their own pages, MP3 to text for recorder and podcast files and M4A to text for iPhone voice memos. Video files work directly, so there is no need to extract the sound first.
Transcript and subtitle formats
| Output | What it contains | Open it in |
|---|---|---|
| Plain text (TXT) | Paragraphs of text, optional timestamps and speaker labels | Any text editor, notes app or email |
| Word (DOCX) | Formatted transcript ready to edit and share | Word, Google Docs, Pages, LibreOffice |
| A fixed copy of the transcript for records | Any PDF reader or browser | |
| Subtitles (SRT) | Numbered captions with start and end times | YouTube, video editors, VLC |
| Web subtitles (VTT) | WebVTT captions, the W3C format used by HTML5 video | Web players, YouTube, course platforms |
| JSON (word timings) | Every word with its start and end time | Scripts, search indexes, editing tools |
Timestamps, speakers and other transcript options
- Timestamps
- Timestamps put the time, such as [01:15], in front of each paragraph of TXT, Word and PDF transcripts so you can find the moment in the recording.
- Speaker labels
- Speaker labels mark who is talking as Speaker 1, Speaker 2 and so on, for interviews, meetings and podcasts. Replace the labels with names afterwards.
- Punctuation and fillers
- Punctuation is on by default. Remove filler words drops um, uh and similar hesitations.
- Characters per line
- For SRT and VTT, Characters per line splits long captions, 42 by default, the limit Netflix's subtitle guidelines use for most Latin-script languages; anything from 20 to 80 is allowed.
- Part to transcribe
- Start at and End at limit the work to one section, in seconds (75) or minutes:seconds (1:15).
Online audio to text converter limits
- Recordings per batch
- 20
- Per audio or video file
- 95 MB
- Spoken languages plus auto-detect
- 33
- Until files and transcripts are deleted
- 30 min
Getting a more accurate transcript
Many words are wrong or missing
Background noise, distance from the microphone and people talking over each other cause most errors. Record closer to the speaker where you can, and check that Spoken language is right.
The transcript is in the wrong language
Auto-detect heard music or another language in the first 30 seconds. Choose the Spoken language yourself and transcribe again.
Subtitles are too long to read on screen
Lower Characters per line to 32–37 for phone viewers, so each caption splits into shorter lines.
Names and technical terms are misspelled
Rare words are hard for any model. Search and replace them in the Word or text output once; they are usually misspelled the same way throughout.
Audio to text questions
Can I transcribe audio to text free without signing up?
Yes. Upload the recording and press Transcribe; there is no account, trial or credit card involved. Files are deleted automatically 30 minutes after processing, so download the transcript when it is ready.
What makes one audio to text converter better than another?
For a fair comparison, look at accuracy on your own kind of recording, the languages offered, the output formats and what happens to your file afterwards. Here you get 33 languages plus auto-detect, six outputs including subtitles, speaker labels and timestamps, and automatic deletion after 30 minutes.
Which languages can be transcribed?
33, chosen under Spoken language: among them English, Spanish, French, German, Portuguese, Italian, Russian, Arabic, Turkish, Hindi, Bengali, Tamil, Telugu, Marathi, Gujarati, Urdu, Japanese, Korean, Chinese, Indonesian, Vietnamese, Thai, Polish, Ukrainian, Dutch, Swedish, Greek and Hebrew. Auto-detect picks one language per file from the first 30 seconds.
Can I make subtitles for a video from its audio?
Yes. Upload the video itself, choose SRT or VTT under Output, and you get a timed subtitle file to load into YouTube, a video editor or an HTML5 player. To change the subtitle format later, use the subtitle converter.
How do I transcribe a 3-hour recording?
Files are limited to 95 MB, which is roughly 1 hour 40 minutes of 128 kbps MP3. Split the work with Start at and End at, or make the file smaller first: speech converted to 64 kbps mono with compress audio fits about 3.5 hours under the limit.
Can the transcript be translated into English?
No, the transcript is written in the language that is spoken. Transcribe first, then paste the text into a translation service; translating a clean transcript gives better results than translating straight from audio, and you keep the original wording for checking quotes.
Is an MP3 to text converter different from a video transcriber?
Only in what you upload. The same speech recognition runs on the soundtrack of an MP4 or MOV and on an MP3 or M4A. MP3 to text and M4A to text are presets of this tool for those file types.
Will um and uh appear in the transcript?
They do by default, because a verbatim record is sometimes needed, for example in research interviews. Turn on Remove filler words for a cleaner transcript meant for reading, quoting or publishing.
All Audio to Text tools
Pick the exact target you were asked for. Every page below is the same tool with different defaults.
- M4A to Text ConverterTurn iPhone Voice Memos and other M4A recordings into text with timestamps and speaker labels. 33 languages, TXT, Word, PDF or subtitles, no account.
- MP3 to Text ConverterTranscribe an MP3 to text in 33 languages. Download plain text, Word, PDF or SRT and VTT subtitles, with optional timestamps and speaker labels.