Transcribe audio to text — AI accurate, with built-in subtitle generator

×

Drag and drop files here or click to select files
mp3, wav, ogg, opus, aac, m4a, flac, amr, aiff, aif, 3gp, webm, mp4, mov, mkv, wmv, avi

+Add more files

Upload File
Convert Youtube
File name Date Duration Status
×

Export


Formats

Configure Export

Transcribe audio to text and convert to SRT/VTT subtitles in one upload — speaker-labelled, timestamped, ready for any video editor.

95–98% AI accuracy Transcript + SRT/VTT subtitles 3-day retention · no training Free 10 min · no signup · no credit card

How to transcribe audio to text — 3 steps

Upload, let the AI work, then tune the export to your workflow.

1

Upload audio or video

Drag & drop MP3, WAV, MP4 — or paste a YouTube link. Files up to 1 GB and 3 hours.

2

AI transcribes

Our model converts speech to text with 95–98% accuracy, timestamps every line, and labels speakers.

3

Configure & export

Tune paragraph rhythm and which timestamps appear, then ship as TXT for writers, DOCX for review, or SRT/VTT subtitles for any video player.

Privacy and data handling — straight talk

If you're transcribing a confidential interview, medical session, legal recording, or internal meeting, here's exactly how we handle the file. No marketing flourish.

Encrypted in transit

Every audio file you upload travels encrypted, and so does the resulting transcript when you fetch it back. SSL/TLS end-to-end.

Auto-deleted after 3 days

Your audio file and its transcript are wiped from our storage 3 days after upload. Need to keep a copy? Download it within that window or set a reminder.

No training on your data

Your audio recordings stay yours. They don't enter any training pipeline. The model that transcribes them is pre-trained and frozen — your file is processed once and forgotten.

GDPR-aligned

EU users have the standard rights — copy, deletion, portability — exercisable from your account or via support. Each upload sits at a private URL keyed to the account that created it.

Your content stays private, encrypted, and entirely under your control.

Supported formats

Audio, video, and YouTube on the way in — every common transcript format on the way out.

Audio in
MP3WAVOGGOPUSAACM4AFLACAMRAIFF3GPWEBM
Video in
MP4MOVMKVWMVAVIWEBM
URL in
YouTubeYouTube Shorts
Transcript out
TXTDOCXPDFSRTVTTCSVClipboard
Demo · not your file

From audio file to ready-to-drop subtitles

Drop audio in — get back a clean transcript and a subtitle file segmented to caption-friendly line lengths. The 38-second sample below converts to SRT and VTT (highlighted) as well as TXT, DOCX, PDF, and CSV. Same shape of output your file will produce.

Audio in · 0:38 Tech-journalism interview
Speaker 1 Speaker 2
Transcribe & export ↓
.srt · Subtitle file
1 00:00:01,200 --> 00:00:04,500 [Speaker 1] What got you started in tech journalism? 2 00:00:05,100 --> 00:00:13,800 [Speaker 2] Honestly, by accident. I was covering...
.vtt · Subtitle file
WEBVTT 00:00:01.200 --> 00:00:04.500 <v Speaker 1>What got you started in tech journalism? 00:00:05.100 --> 00:00:13.800 <v Speaker 2>Honestly, by accident...
.txt
[00:01] Speaker 1: What got you started in tech journalism? [00:05] Speaker 2: Honestly, by accident. I was covering city hall, and one source kept saying things I had to translate for readers — that was the click. [00:14] Speaker 1: How long until you knew it was the beat?
.docx

Speaker 1 · 00:01
What got you started in tech journalism?

Speaker 2 · 00:05
Honestly, by accident. I was covering city hall…

.pdf

Tech-journalism interview
Sample transcript · 0:38

Speaker 1 · 00:01
What got you started in tech journalism?

.csv
start,end,speaker,text 00:01,00:04,Speaker 1,What got you started in tech journalism? 00:05,00:13,Speaker 2,Honestly by accident...

Effectively a built-in subtitle generator: SRT and VTT files are pre-segmented to caption-friendly line lengths (≤ 42 chars per line) — drop straight into Premiere, DaVinci, Final Cut, CapCut, or YouTube Studio. Or tune all output controls first.

Configure your output the way you need it

Most transcription tools dump a single wall of text. Ours splits the transcript by speaker, by pause length, and by paragraph rhythm — tuneable to your downstream tool.

Paragraph length

Adjust how long each paragraph is

Auto-detect or fix every paragraph at 1, 2, 3, 4, or 8 lines. Useful when you're pasting into a doc that has its own preferred rhythm.

Auto1 line2 lines3 lines4 lines8 lines
Paragraph breaks

Tune where new paragraphs start

The AI starts a new paragraph after a pause. Adjust the pause length to taste — shorter for fast speech, longer for measured monologue.

500 ms700 ms (default)1500 mscustom
Timestamps

Show timestamps where you want them

Per paragraph for skim review, per phrase for legal-cite work, both for full audit trail, or off for clean publishable prose.

ParagraphsPhrasesBothOff
Speakers

Name speakers, or merge consecutive turns

Auto-labelled Speaker 1 / Speaker 2. Rename in the editor to match the panelists, hosts, or interview subjects you uploaded.

Speaker namesMerge by speakerHide
Plain text mode

Strip everything but the words

One toggle collapses the transcript to publishable prose — ready for a writer, an LLM summarizer, or to paste into a CMS draft.

Plain text mode
Clipboard

Copy without downloading a file

Skip the file step. Paste the configured transcript straight into Notion, Google Docs, or your CMS — already in the right shape.

Copy to clipboard

Languages we transcribe with near-native accuracy

These are the languages where the model delivers consistently strong results. Auto-detect picks the right one; mixed-language clips work too.

  • English
  • Spanish
  • Mandarin Chinese
  • Portuguese
  • German
  • French
  • Italian
  • Russian
  • Japanese
  • Korean
  • Hindi
  • Arabic

If your audio is in a less common language, run a 60-second sample on the free tier first.

Built for the way you work

One transcript engine, every workflow that needs words from sound.

Journalists & researchers

Transcribe audio to text from interviews and field recordings — speaker-labelled output and a transcript generator built for fast quote-pulling.

Educators & students

Turn lectures and seminars into study notes. Add timestamps and skim instead of re-listening.

Podcasters & creators

An audio to text converter that doubles as a show-note generator: feed in an MP3, get blog repurposes, episode summaries, and chapter cues.

Subtitle creators

Use the built-in subtitle generator to produce SRT and VTT files for YouTube, TikTok, and any video player.

Legal & compliance

Transcribe depositions, hearings, and meetings with timestamps for line-cite review.

Teams & meetings

Drop a meeting recording, get an action-item transcript ready to paste into your doc tool.

Free tier — try before you commit

Test transcription quality on your own audio. No credit card. Top up only when you need more minutes.

Free

10 minutes / month Full features. No signup. No watermark. No subscription.

Top-up

From $4.99 Single payment for a minute pack. Minutes never expire — no monthly reset, no subscription.
See plans

Transcription FAQ

The questions we hear most from new users — answered straight.

How accurate is the transcription, really?
95–98% on clean speech. Heavy accents, background noise, overlapping voices, or compressed phone audio pull accuracy down — sometimes well below 95%. The hero number is the ceiling, not the floor. For anything you'll publish or cite, plan a review pass in the editor.
How long does transcription take?
It depends on file length and current load. Most files complete within several minutes per hour of audio; busy periods or longer uploads take longer. You'll see live progress, and you can leave the tab — we keep working in the background.
What happens if my audio is poor quality?
The transcript will still come back, but expect mistakes. Background noise, thick accents, two people talking at once — these are where AI struggles. Open the built-in editor, scrub the audio while you read, fix the lines that matter, then export. The 3-day retention gives you a window to do this without rushing.
Beyond the listed languages, does it work?
Often, yes — but quality varies. Less-common languages and regional dialects may transcribe with lower accuracy than the listed top languages. We recommend running a short sample on the free tier first to see whether the result is usable for your specific source.
Can I share a transcript with someone else?
Yes. Each transcript lives at a unique URL — share the link with people who should see it, or just download and email the file. Remember the page auto-deletes after 3 days, so collaborators should grab a copy if they need long-term access.

Other transcription tools

We use cookies to ensure you get the best experience on our website. Learn more: Privacy Policy

Accept Cookies