Board and leadership meetings
Upload the recording and get it back with every speaker separated and every word timed. Name a voice once, and later uploads label it for you.
Speaker labels · voiceprints from Pro up
DOVEV Transcription
Every word,
exactly as spoken.
DOVEV turns audio and video into precise, timestamped transcripts in 99 languages, then translates them into 88 and exports subtitles that fit.
board-interview.mp4Transcribing
David Whitmore00:03
So let me introduce everyone before we start recording properly.
Claire Bennett00:09
Thanks David. I'll be taking notes and I'll share them by Thursday.
00:14 / 42:07
Bring audio in from anywhere. Leave with the format you need.
YouTube
Zoom
Google Drive
Dropbox
Zapier
Any public URL
MP3 · WAV · M4A
MP4 · MOV · MKV
DOCX · PDF · CSV
SRT · VTT
5× faster than real time
A one-hour recording comes back in about twelve minutes, already punctuated, attributed and timestamped, not a wall of text you still have to clean up.
Typing it yourself
4 to 6 hrs
Play, pause, rewind, retype. Then label the speakers, then scrub for every timestamp, then do it again for the second language.
With DOVEV
~12 min
- Punctuated and paragraphed
- Speakers separated on paid plans
- Word-level timestamps
99%
Accuracy on clear audio
99
Languages transcribed
Word
Level timestamps
256-bit
Encryption at rest
Three independent actions,
not a locked pipeline.
Transcribe first. Then translate, or subtitle, or neither. Each runs only when you ask, and only charges for the work you trigger.
Timed to the word
Timed to the word, not the paragraph
Every word carries its own start and end. Click one and the player jumps there; edit one and the timing survives. That is what makes captions land on the right frame.
transcript · English
Dr. Helen Cartwright12:04
The second cohort showed a measurable improvement across both endpoints.
Word start 12:06.4 → player seeks here
Knows who spoke
Knows who spoke, and keeps them straight
Speaker separation runs automatically, so interviews and board calls read as a conversation instead of a wall of text. Rename a speaker once and every turn updates.
speakers · auto-detected
Dr. Helen Cartwright12:04
The second cohort showed improvement.
Marcus Bell12:11
Across both endpoints, or only the primary?
Dr. Helen Cartwright12:14
Both. I'll send the table over.
Dr. Helen Cartwright Marcus BellRename once, applies everywhere
Translate on demand
Translate without losing the timeline
Turn a finished transcript into any of 88 languages. Each language becomes a tab on the same timeline, with the same speakers and the same timestamps, so subtitles still fit.
translation · timeline preserved
EnglishEspañolहिन्दी+84
12:04
The second cohort showed improvement.
La segunda cohorte mostró una mejora.
12:11
Across both endpoints?
¿En ambos criterios de valoración?
12:14
Both. I'll send the table over.
Ambos. Enviaré la tabla.
Speech goes in messy.
Text comes out clean.
Fillers, false starts and repeated words are removed on the way through. Punctuation, paragraphs and speaker turns are added. You get something you can send.
standup-2026-08-07.m4aDrop a file to start
Audio or video, up to 2 GB, or paste a link
Listening
No plugins, no setup Shared workspace balance Auto language detection
The details that decide
whether you use it twice.
Language: English
99 languages in, 88 back out
DOVEV detects the spoken language on its own and transcribes it. A finished transcript can then be translated into any of 88 targets, each becoming a tab on the same timeline rather than a separate file.
It learns your vocabulary
Client names, drug names, internal jargon, acronyms. Add a term to your dictionary once and it goes to the engine as a hint on every future transcript, steering it toward your spelling instead of a phonetic guess.
Subtitles that fit the frame
SRT and VTT generated from stored word timings, with cue lengths that are actually readable. Captions in the source language are always free.
Ask a recording what happened
Every finished file gets an AI summary — topics, decisions, action items — and a panel you can question in plain language. Answers come from that transcript alone, so when it isn't in the recording, it says so instead of inventing one.
Language: English
99 languages in, 88 back out
DOVEV detects the spoken language on its own and transcribes it. A finished transcript can then be translated into any of 88 targets, each becoming a tab on the same timeline rather than a separate file.
CartwrighttirzepatideQ3 OKRs
Add a new wordIt learns your vocabulary
Client names, drug names, internal jargon, acronyms. Add a term to your dictionary once and it goes to the engine as a hint on every future transcript, steering it toward your spelling instead of a phonetic guess.
We’ll have a firm timeline
by end of day Thursday.
00:12,480 → 00:15,120SRTVTT
Subtitles that fit the frame
SRT and VTT generated from stored word timings, with cue lengths that are actually readable. Captions in the source language are always free.
Ask DOVEV
What did we decide about the terms page?
Legal signs off Thursday, then it ships with the release. Priya owns the copy.
What decisions were made?List any action items.
Ask a recording what happened
Every finished file gets an AI summary — topics, decisions, action items — and a panel you can question in plain language. Answers come from that transcript alone, so when it isn't in the recording, it says so instead of inventing one.
Privacy
Your recordings
stay yours.
Encrypted in transit and at rest. Never sold, and never used by DOVEV to train a model.
board-interview.mp4AES-256
00:03David Whitmore
So let me introduce everyone before we start.
00:09Claire Bennett
Thanks David. I'll share my notes by Thursday.
00:21David Whitmore
Keep the board figures out of the summary.
On your device
Encrypted in transit and at rest
HTTPS on every connection, AES-256 on every stored file.
Never used to train AI
DOVEV never uses your recordings or transcripts to train a model.
Scoped to your account
Row-level security scopes every file to its owner, and media is only served through links that expire.
Deleted for good
Delete a file and it is purged from DOVEV's storage and database after 7 days in Trash, or at once if you empty it.
The same hour of audio,
two very different afternoons.
By hand
DOVEV
Time for a 60-minute recording
4 to 6 hours of typing
Around 12 minutes
Who said what
You label every turn by ear
Speakers labelled automatically on paid plans
Timestamps
Scrubbed and typed by hand
Every word carries its own timing
Second language
A translator and another invoice
88 translation languages, one timeline
Subtitle files
Retimed manually per format
SRT and VTT in one click
Cost model
Per-minute agency rates
One pooled balance, prorated to the second
In the workflow
What people do
with the time back.
Publishing in more than one language
Translate a finished transcript into any of 88 languages. Each one keeps the original timings, so its subtitles still line up.
Translation · SRT and VTT export
Research with specialist vocabulary
Add drug names, product names and jargon to your custom dictionary. They reach the engine as hints, so the first draft needs fewer fixes.
Custom dictionary · medical mode on Business
Freelance production
One balance covers transcription and translation, charged by the second of media. Top up by the hour when a big job lands.
One combined balance
Interviews and reporting
Click any word and playback jumps to it, so checking a long interview means replaying the parts in doubt, not the whole thing.
Word-synced editor
Review with outside parties
Share a transcript with one person at a time, as view, comment or edit access, limited to the languages you choose. Revoke it whenever you like.
Per-person sharing
FAQ
Good questions.
Questions
Answer
How accurate is it, really?
Up to 99% on clear audio with a decent microphone. Accuracy drops with heavy crosstalk, distant mics and background noise. That is why every transcript opens in an editor with the audio synced word by word, so correcting the last few percent takes seconds rather than a second pass.
Still stuck? Talk to us.
Start transcribing.
Your first 30 minutes are free. See what each action costs before you run it. You never pay for work you didn’t ask for.
No credit card required · Cancel anytime