quarterly-review.mp4
00:14MayaRevenue grew fifteen percent, and the retention metrics are green.
00:31DanielGood. What are we committing to for Q2?
00:38PriyaConcrete steps by Friday. I'll own the rollout plan.
99 languages · Unlimited on paid plans
Turn speech into text, summaries, and subtitles in minutes
Upload audio or video and get an accurate transcript with speakers separated, a summary with the actual decisions, and subtitles ready to ship. No per-feature upcharges.
- No credit card to start
- GDPR · EU hosting
- Never trained on your data
How it works
Three steps. None of them yours.
- 01
Drop the file in
Audio or video, any common format, up to 5 GB. A link to a recording works too.
- 02
The engine does the work
Language detection, transcription, speaker separation, punctuation, and timecodes. Nothing to configure.
- 03
Take what you need
Clean text, a dialogue split by speaker, an AI summary with decisions, or subtitles — export in a click.
Everything included
The features other tools sell as add-ons are just part of the plan.
Diarization, summaries, and subtitles are often billed separately elsewhere. Here they are included on every paid plan, from the first minute.
Speaker diarization
Up to 6 speakers separated automatically, with labels you can rename.
AI summaries
Decisions, action items, and owners — not a paraphrase of the whole call.
Subtitles
SRT, VTT, and ASS, or burned straight into the MP4 without a watermark.
Word-level timecodes
Every word carries its own timestamp, so editing stays in sync with audio.
99 languages
Automatic detection, including mixed-language recordings.
Export anywhere
Word, PDF, TXT, and subtitle formats. Your transcript is not locked in.
Who it's for
Who saves the most time
Teams that live in meetings
Decisions and owners captured without anyone playing scribe.
Researchers
Interviews transcribed verbatim, with speakers separated for coding.
Journalists
Quotes with exact timecodes you can defend against the recording.
Podcasters
Show notes, chapters, and subtitles from the same upload.
Students
Lectures turned into searchable notes you can actually revise from.
Sales teams
Call reviews and follow-ups written while the deal is still warm.
Beyond a single file
Not just transcribing an upload.
An AI notetaker that joins the call
Invite the bot to Zoom, Google Meet, or Teams. It joins, records each participant on a separate track, and posts the transcript and summary when the call ends.
Meet the notetaker →An API you can build on
Send a file, get structured JSON back: text, speakers, word-level timings, and a summary. Diarization and subtitles are in the rate, not extra.
Read the API docs →What we can prove
The numbers we'll stand behind.
NUMU PRO is new, so there are no customer quotes here yet — we'd rather show the engine's actual characteristics than invent testimonials.
- 99%accuracy on clean audio
- Measured on clear recordings. Noisy rooms and heavy accents land lower — we say so rather than promise a number we can't hold.
- 99languages supported
- Detected automatically, including recordings that switch language mid-sentence.
- 6speakers separated
- Diarization labels each voice so a meeting reads as a dialogue, not a wall of text.
- EUhosting and data residency
- GDPR compliant, DPA available, and your recordings are never used to train models.
Pricing
Clear pricing. Free to start.
Every paid plan includes unlimited transcription and every feature. What changes is team size, API access, and support.
Free
Try the engine on your own files
$0/mo$0/moFree forever
Starter
Unlimited transcription for one person
$8.50/mo$4.25/moBilled monthly$51 billed yearly
- Most popular
Pro
For small teams that live in meetings
$19/mo$12.54/moBilled monthly$150.48 billed yearly
Business
Team workspace with SSO and roles
$59/mo$38.94/moBilled monthly$467.28 billed yearly
Enterprise
On-premise, air-gapped, or custom volume
CustomPriced to your volume
* Unlimited transcription on all paid plans, subject to a fair-use clause in the terms of service.
FAQ
Common questions
How accurate is the transcription?
Around 99% on clear recordings — a good microphone, one person speaking at a time. Background noise, strong accents, and heavy crosstalk lower that, as they do for every engine. You get 60 free minutes precisely so you can measure it on your own audio instead of taking our word for it.
What does "unlimited" actually mean?
Unlimited transcription minutes on every paid plan. There is a fair-use clause in the terms to stop abuse at extreme volume — it exists for resellers pushing tens of thousands of hours, not for normal work. If your usage is ever a problem, we contact you before doing anything.
Which languages are supported?
99 languages, detected automatically. Recordings that switch language mid-sentence are handled too, which matters for multilingual teams.
Do you train models on my recordings?
No. Your audio and transcripts are never used to train models, on any plan including Free. Data is hosted in the EU, we are GDPR compliant, and a DPA is available.
Can I get subtitles, not just text?
Yes. SRT, VTT, and ASS files on every paid plan, or subtitles burned directly into the MP4 without a watermark. The Free plan exports SRT and adds a watermark on burn-in.
Is there an API?
Yes, from the Pro plan up. Send a file, get JSON back with text, speakers, word-level timings, and a summary. Diarization and subtitles are included in the per-minute rate rather than billed as extras.
What happens when I hit my Free limit?
Nothing breaks and nothing is deleted. You keep access to everything already transcribed, and you can either wait for the monthly reset, buy minutes as you go from $0.10, or move to a paid plan.
Upload your first file now.
Sixty minutes free every month, no card required. See what the transcript looks like on your own audio before you decide anything.