Speaker identification

Speaker Identification (Diarization): Know Who Said What

Cockatoo finds every speaker on its own and labels who said what, so you can quote the right person without replaying the audio. Speaker labels come with V4-Pro, our most accurate mode, and free accounts get V4-Pro previews every week to try it on their own recordings.

Automatic Speaker Recognition Labels Who Said What in Your Transcript

Upload your audio or video file and get back a transcript with every speaker labeled and separated automatically.

No manual tagging. No replaying the same ten seconds to figure out who spoke. No hours lost.

  • Speakers detected automatically
  • Rename once, every line updates
  • 36 languages
  • No credit card
Cockatoo transcript editor with speaker labels and timestamps
A transcript with timestamps in the Cockatoo iPhone app
Drag and drop
MP3MP4M4AMOVAACWAVOGGOPUSMPEGWMV
OR
Speed and accuracy

Fast Enough to Keep Up, Accurate Enough to Trust

120xreal-time speed in V4-Flash, at most
~20xreal-time speed in V4-Pro, with speaker labels
36transcription languages

Transcribe a One-Hour File in About 30 Seconds

Cockatoo transcribes at up to 120x real-time in the fastest mode, V4-Flash, so you get results while the context is still fresh. V4-Pro, the mode that labels speakers, runs at about 20x real time.

Upload your recording, grab a coffee, and your labeled transcript is waiting when you sit back down.

120xreal-time speed in V4-Flash, at most

Blazing Speed. Incredible Accuracy.

Fast transcription is worthless if you spend an hour fixing mistakes. High accuracy does not help if it takes all afternoon.

Cockatoo delivers both, so you can trust the transcript and move on to the work that matters. Accuracy varies by language and recording quality, so check the published benchmarks.

Transcription mode

V4-FlashUp to 120x real timeFree, unlimited
V4-ProIndustry-leading accuracyAbout 20x real time
focus-group-2.mp3Transcribe
Edit and export

Export Speaker-Labeled Transcripts to the Formats You Already Use

Labeled, corrected and named, then out the door in the format you need.

Export as SRT Subtitle File Alongside DOCX, PDF, and TXT

Download your speaker-labeled transcript in the format your video editor, word processor, or publishing tool expects. VTT, CSV and JSON are there too.

SRT files carry the speaker tags and timestamps, so subtitles stay synced and attributed.

Cockatoo export menu with DOCX, PDF, TXT, SRT and more
Proof

Trusted by Professionals Who Check Every Word

What Our Users Appreciate: Accuracy on Difficult Accents and Dialects, Raw Speed, and Ease of Use

Our users mention regional accents, mixed-language conversations, and multi-speaker recordings as scenarios where Cockatoo delivered when other tools failed.

Cockatoo V4-Pro on FLEURSError rate · lower is better
  • English3.9% WER
  • Spanish1.0% WER
  • German2.4% WER
  • French7.8% WER
  • Japanese6.5% CER

Accuracy varies by language and recording quality. Updated October 27, 2026. Full results and method

Published Per-Language Accuracy Tests Covering 36 Languages

Cockatoo publishes accuracy test results for 36 languages: the same public test set researchers use to compare transcription engines, plus real-world audio examples.

You can audit the claims yourself instead of trusting a marketing percentage with no evidence behind it.

Detected languageالعربية
  • English
  • Español
  • العربية
  • 日本語
  • עברית
  • Deutsch
  • فارسی
  • Français
  • اردو

00:14 مرحبًا بكم في المحاضرة الرابعة

Mixed-Language Files Supported, Auto-Detect Runs Before Transcription

Upload a recording where speakers switch between languages, and Cockatoo transcribes it in one pass.

In V4-Flash, auto-detect runs before transcription. In V4-Pro, you choose the language at upload. No need to split the file into separate uploads.

“Simple to use, very accurate - can identify between speakers and quicker than most in identifying franco arabic conversation.”
Dina S.CanadaRead review
“Cockatoo separated speakers and sentences, with timestamps, making it way easier to read.”
Heidi C.CanadaRead review
“This transcriber is incredibly accurate, and not bothered by all the different accents i present it with. Highly recommended.”
Ian A.AustraliaRead review
Plans

Try Speaker Identification Free, No Credit Card Required

One account, two modes: V4-Flash for unlimited free transcripts, V4-Pro for speaker labels and industry-leading accuracy.

How V4-Pro and V4-Flash compare
FeatureV4-ProV4-Flash
Best forKnowing who said what, with speaker labels detected automaticallyFast, unlimited transcripts
SpeedAbout 20x real timeUp to 120x real time
Who gets itWhole recordings on paid plans, weekly previews on FreeUnlimited on every plan, including Free
LanguageChoose it at uploadDetected for you, or choose it
EditingCorrect any line, and rename or merge speakers in the browser editorCorrect any line in the browser editor
Export formatsDOCX, PDF, TXT, SRT, VTT, CSV, JSON, plus Adobe Premiere markers on paid plansDOCX, PDF, TXT, SRT, VTT, CSV, JSON

Plans and prices change, so check the pricing page for current details.

FreeUnlimited full-length transcripts in V4-Flash$0
ProWhole recordings in V4-Pro, priority processing, unlimited file length
Max5x or 10x the Pro V4-Pro allowance, for backlogsFrom $100/mo
TeamA shared workspace for 3 or more people
See current prices

No Monthly Hour Cap on V4-Flash

Upload as many hours as you need in V4-Flash. No surprise overage fees. No throttling when you hit an invisible limit.

For speaker labels, use your weekly V4-Pro previews on Free, or transcribe whole recordings in V4-Pro on Pro.

30% off

For students, academics and non-profits, on paid plans

Request the discount

Discounts for Students, Academics and Non-Profits

Recording research interviews or focus groups? Students, academics and non-profits can request a discount on paid plans from the education page.

FAQ

Common Questions About Speaker Identification

Short answers about speaker labels, plans and editing.

Visit the help center

Cockatoo publishes accuracy tests for 36 languages: the same public test set researchers use to compare transcription engines, plus real-world audio examples you can listen to and compare.

Find out who said what in your next recording

Get your full transcript free, with no daily limit. Try speaker labels with your weekly V4-Pro previews.

Start Transcribing Free