Saw an interesting video from @andreijikh about South Korea’s AI bubble having popped and couldn’t quite grasp some of the context even after rewatching a few times - In comes transcribe.so with the save, giving me a full transcript and letting me ask questions 🙌🔥
ProfessionalsStudentsCreators
Find who said what and exactly when they said it.
Connect once, then paste a link into ChatGPT, Claude, or any agent and ask for chapters, highlights, quotes with who said it and when, or subtitles. The hours of work after a recording, done in minutes. Upload files in the web app.
No credit card required.
7-day money-back guarantee
On your first charge, if you have used under 10 hours of transcription.
Speaker labels·Word-level timestamps·52 languages and dialects·Encrypted storage·API, MCP, CLI
The full-length interview with Elon Musk | The Economist
The Economist
Try this real transcript
Who said what, when
Contents✍️Human written like chapters
20 chapters · 60 sections
1The Age of Abundance is Coming
2Why You Won't Control Superintelligence
3AI Doom is Inevitable (Eventually)
4How Rivals Keep AI Honest
5China's AI Rise Changes Everything
Ask this video
🎯Get who said what, whenAnswer
Sooner than most people expect: Musk guesses AI may exceed the sum of human intelligence in roughly five years, and be far greater than it within ten . He says it is unlikely humans will still be in control ten years from now, comparing the gap to the one between humans and chimpanzees . And he admits progress looks so inexorable that he could not stop it even if he wanted to .
Command Palette
Search for a command to run...
Real posts from X
People use this to find who said what
Thanks @shsunmoon, I used @transcribeso twice for tricky Arabic YouTube videos, and the results were really good! It gave me full Arabic transcripts and kept them in my account for later retrieval so I could share them with ChatGPT instead of watching the full videos!
So I have been using handy and its really handy hahaha. it failed to transcribe a 10 min voice memo. That kinda sucked. I tried out transcribe.so and it did an amazing job. it can also handle hours of video&audio.
The real cost of long recordings
You remember who said it. You just can't get back to the moment.
A client agrees to a scope change. A professor finally explains the concept. A guest drops the line you want to quote. The recording knows exactly who said it and exactly when, and it is not telling. So you scrub, overshoot, rewind, and settle for "I think it was somewhere in the middle."
Per long recording
3-hour recording with one moment you need back
+ 20 to 40 min scrubbing to find who said it
+ 30 to 60 min re-listening at 1.5x
+ A summary with no names and no timestamps
+ ∞ min wondering which recording it was in
= An evening, gone.
There's a faster way.
See it in under a minute
Most people don't need another transcript
They need to do something with the recording. Drop a YouTube link into ChatGPT or Claude and ask for chapters, cited answers, short clips, or subtitles.
See it for your work
Your kind of audio, transcribed and answered
A YouTube deep-dive, a team meeting, a full-length interview — each one below is a real recording processed by the real pipeline. Click around.
Upload a meeting recording and get a recap with decisions you can jump back to.
How it works
From recording to useful in three steps
Real screenshots from a real transcript. This is what you get.
- 1
Add anything
Paste a link or upload a file. Meetings, lectures, podcasts, videos, and voice notes all work the same way.
Time to start: under a minute
- 2
Get a structured transcript
Every transcript comes back with speaker labels, word-level timestamps, summaries, and chapters that read like a human wrote them, not like an auto-generated outline.
Hours of audio become minutes of reading
- 3
Use the recording
Search it, ask questions with cited answers, jump to the exact moment, and export notes or subtitles when you need them.
Elon Musk0:00Elon Musk on Building the FutureInterviewer31:57Future Work as Artisanal and Human-CentricEvery answer names the speaker and the moment.
No credit card required.
Built for where you post
Same words. Captions that fit each platform.
Every word carries its own timestamp, so a preset can decide how many words sit on screen at once. Each tab loops a short excerpt of a real video from our demo library with captions generated from its transcript, word by word. Sound is off until you turn it on. Switch the caption style to compare platforms.
9:28.49Every word carries its own timestamp. Hover or tap any word.
From PowerfulJRE, “Joe Rogan Experience #1470 - Elon Musk”, transcribed by transcribe.so
Caption style
Two-line captions and description-ready chapters
YouTube preset: 2 lines, 38 characters per line, up to 6 seconds per cue
- Captions export as SRT or VTT and pass the YouTube caption upload as-is
- Chapters follow the rules YouTube enforces: start at 0:00, at least 3, 10 seconds apart
- Pinned-comment timestamps and show notes come from the same transcript, one click each
- Clip ideas point at the 3 moments worth cutting into Shorts
A note from the maker
Hey, I’m Seunghun 👋
I left Spotify in 2023 to build a startup around finding important moments in long-form content.
The AI wasn’t good enough yet, and the business failed.
In 2025, I realised the technology had finally caught up. So I quit my job and built transcribe.so.
I’ve always enjoyed listening to insightful conversations. But finding something again meant replaying hours of content.
I had to build a better way to find who said what and when.
What you can ask for
The work after the recording, done for you
Chapters that read like you wrote them
Chapters for my YouTube description
A timestamped chapter list ready to paste into a description or show notes. No scrubbing the whole video to write it.
Highlights, ready to share
The best moments for a Reel
The moments worth sharing, as a short outline or timestamped clip ideas for Shorts, Reels, and TikTok. No rewatching to find them.
Quotes with who said it and when
What did the guest say about pricing?
The speaker's exact words and the moment they said them, with a link that opens playback right there. No re-listening at 1.5x.
Subtitles your way
SRT subtitles for YouTube
SRT or VTT files formatted for YouTube, Reels, TikTok, Netflix-style, podcast, or broadcast. Translated and bilingual SRT in the web app on Pro and above. No hand-timing captions.
Search and cited answers
Ask the recording. Verify the answer.
A summary tells you what a recording was about. Asking tells you who said what and exactly when, and every answer opens at the moment it came from.
A real question on a real transcript
Will AI take your job, according to Elon Musk?
He expects a bumpy road: AI will be able to do any job better than a person, starting with anything done at a computer or on a phone, and he says it already writes software better than about ninety percent of professional engineers. Next comes 'Stockfish level', the chess engine that could beat Magnus Carlsen running on a phone, and he says that applies to everything. His conclusion is that work becomes optional, like growing your own tomatoes when the store's are better.
- Answers come from the recording itself, not from general knowledge
- Every answer cites who said it and the exact moment they did
- Click a citation and playback opens at that exact second
- Ask across one file or your whole library at once
Quality and languages
A transcript you can use without spending an hour fixing it.
A transcript is only useful if you can trust it. Ours come back ready to read, search, and quote.
- Clean punctuation, casing, and paragraphs, not a wall of lowercase text
- Speaker labels on multi-speaker audio
- Word-level timestamps you can rely on for subtitles and clips
- Accuracy that holds up outside English, with per-language error-rate bands from published benchmarks
Fewer errors
published same-split benchmarks
1.5x to 2.3x fewer errors than Whisper large-v3 in published same-split comparisons.
52
languages and dialects supported
Benchmarked across 52 languages and dialects, including Cantonese, Mandarin, Japanese, Korean, Arabic, Hindi.
Stop scrubbing. Start asking.
Privacy
Your recordings stay yours.
Your audio and transcripts are processed on infrastructure we operate and by trusted third-party providers acting on our behalf, and those providers do not use your files to train AI models; the details are in our Privacy Policy. We do not use your content to train AI models unless you opt in.
- Encrypted in transit and at rest
- Your content is never sold
- Delete individual recordings or your whole account anytime
- We do not use your content to train AI models unless you opt in
A library, not a one-off transcript
Every recording becomes part of your memory.
Most tools hand you one transcript and forget it. Here every meeting, lecture, and video joins a library that stays searchable for as long as you keep it.
Search across every transcript
Find a quote, a topic, or what a specific speaker said about the budget, across hours of recordings in seconds.
Ask across multiple recordings
One question can draw on your whole library, with citations back to each source.
Reopen the original moment
Every result links back to playback, so you can hear exactly how it was said.
Keep everything together
Chapters, notes, and exports stay attached to each recording instead of scattering across tools.
For developers
Give your product the same listening engine.
The engine behind this page is an API. Send media, get back structured transcripts with speakers and timestamps.
- Transcribe audio and video with one HTTP call
- Speaker identification and timestamped output built in
- An MCP server for agents, plus an OpenAPI spec for everything else
- One Bearer token; wallet and history stay in sync with the web app
Pricing
Transcribe more without watching the clock.
Start with $3 of free credit. Choose pay-as-you-go at $1/hour for occasional files, or an unlimited plan from $9/mo for regular work. One standard engine handles every file, nothing to pick.
Free
No subscription
$0
- Standard transcription at $1/hour
- All AI features: summaries, chapters, speaker labels, search
- AI Q&A
- Files up to 90 minutes
- Wallet credits never expire
- API and MCP access on every plan
Popular
Pro
Unlimited, for serious work
$19/mo
- Files up to 5 hours
- Unlimited AI Q&A with longer answers
- Email support
Business
Unlimited, built for teams
$49/mo
- Everything in Pro
- 15 parallel API jobs for automation
- Files up to 10 hours
- Priority processing
Choose annual and get 3 months free. Starter and Enterprise plans are on the full pricing page.
Cancel anytime. 7-day money-back guarantee on your first charge on a paid plan, once per account and per payment card, if you have used under 10 hours of transcription since that charge.
FAQ
Before you try transcribe.so
Find who said it. Jump to the exact moment.
Add your first meeting, lecture, podcast, or video and turn it into a transcript you can search, ask, and verify.
No credit card required.