Transcribe audio and video into transcripts you can actually work with.
Upload a meeting, interview, lecture or sermon. Get back an accurate transcript with speaker names, an editor built for fixing it fast, AI notes, and seven export formats.
Free plan: 3 files a day, up to 30 minutes each. No card needed.
The first 8 passages of a 6:12 meeting. Press play or click any word to move the playhead; no audio plays in this demo.
A workspace built for fixing transcripts
Automatic transcription gets you most of the way. The rest is names, numbers and the odd misheard word, so the editor is built around fixing those quickly while you listen.
Speakers
Every passage carries a speaker label. Rename A, B and C once and the whole transcript updates.
Editing
Click a passage and fix it in place. Changes save as you type; merge passages with one action.
Find and replace
Correct a misheard name across the whole transcript at once, case-sensitive or not.
Review
Play along at 0.5x to 2x. The current passage and word stay highlighted, so you always know where you are.
Private by default
Recordings are sensitive. Storage is private, the audio can delete itself on a timer, and the only services that ever see your data are the ones doing the transcription and the notes.
Private storage
Recordings live in a private bucket only your account and the server can read. Media is served through short-lived signed links.
Retention timers
Delete the original audio automatically: right after transcription, after 24 hours, 7 days, 30 days, or never.
Delete audio now
One click removes the media while the transcript stays exactly as it is.
No training on your data
Nothing you upload trains a model, by us or by the services we use to transcribe and summarize.
Outputs you can hand to someone
Turn an hour of talking into the thing people actually read: a summary, a list of decisions, subtitles, or a clean document.
AI notes
Quick summary, detailed summary, action items, meeting minutes, outline, sermon notes, or your own prompt.
Seven export formats
TXT, Markdown, DOCX, PDF, SRT, VTT and JSON, with or without speakers and timestamps.
Copy in one step
Plain text, with speakers, or with timestamps, straight to the clipboard.
Smart titles
A short descriptive title is suggested after transcription. The original filename is always kept.
How it works
Upload a file
MP3, WAV, M4A, AAC, FLAC, OGG, MP4, MOV or WEBM, up to 5 GB. Uploads resume if the connection drops.
We transcribe it
Speaker labels and automatic language detection across 25 languages. Most hour-long recordings finish in a few minutes.
Edit, summarize, export
Fix names, generate notes, and export in the format the next person needs.
Simple pricing
Start free. Upgrade when you need longer files or more of them.
Free
For the occasional recording.
$0
No card needed
- 3 files a day
- Up to 30 minutes per file
- Files up to 500 MB
- Speaker labels, editing, find and replace
- AI notes and seven export formats
- Standard queue
Unlimited
For people who transcribe every week.
$12/month
or $96 a year ($8 a month)
- Unlimited files
- Up to 10 hours per file
- Files up to 5 GB
- 100 hours of transcription a month (fair use)
- Speaker labels, editing, find and replace
- AI notes and seven export formats
- Priority queue
Questions
How accurate are the transcripts?
Transcripts come from AssemblyAI's Universal speech model with speaker labels. With clear audio from a decent microphone, expect a transcript that is ready to share after a quick read-through. Crosstalk, background noise and unusual names lower accuracy, so the editor is built for fixing those spots fast, and you can add key terms before transcribing to help with names and jargon.
Which languages are supported?
25 languages, including English, Spanish, French, German, Portuguese, Japanese, Chinese, Korean, Arabic and Hindi. Leave the language on automatic detection or set it for each upload.
Does it tell speakers apart?
Yes. Speaker identification is on by default and labels every passage. Rename a speaker once and the name updates everywhere, including exports. If you know how many people are talking, set the number before transcribing to improve the split.
What are the file limits?
Free: 3 files a day, up to 30 minutes and 500 MB per file. Unlimited: files up to 10 hours and 5 GB, with 100 hours a month as fair use. Supported formats: MP3, WAV, M4A, AAC, FLAC, OGG, MP4, MOV and WEBM.
Who can see my recordings and transcripts?
Only you. Recordings are stored in a private bucket tied to your account and served through short-lived signed links. To transcribe, AssemblyAI receives a temporary link to the audio. When you run an AI tool, Anthropic or OpenAI receives the transcript text and your instruction. Neither we nor those services train models on your data, and nobody else receives it. You decide how long the audio is kept.
Can I cancel anytime?
Yes. Cancel from Settings and Unlimited stays active until the end of the period you already paid for. There are no refunds for partial months or years. Your account moves back to the free plan and your transcripts stay yours.