kasete
Open app
RECORD · TRANSCRIBE · DONE

Hit record. Talk.Walk away.

Kasete captures every voice and returns a multi-speaker, multi-language transcript as a single Markdown file, ready to drop straight into your AI agent.

Mic or Browser Audio Up to 2 hours Markdown export
9:41● ● ●
Record
12:47
MicrophoneTab audio
Recording… tap to pause
CancelDone
Recent3
Weekly sync — Growth
42:18 · completed
Customer call · Acme
28:05 · completed
1:1 with Dewi
16:40 · processing
weekly-sync.md

You sat through the whole meeting. Now you're scrubbing a recording, retyping quotes, and guessing who said what. All just to give your tools the context they needed all along.

the busywork ends here
How it works

From spoken words to agent-ready text.

Three steps. No note-taking, no copy-paste archaeology, no wondering who said what.

01

Record

One tap. Capture the room through your mic or pull audio straight from a browser tab. Pause, resume, up to two hours.

mic · Browser Audio
02

Transcribe

Kasete diarizes and transcribes automatically. Every line is labelled by speaker and tagged by language.

diarized · multi-language
03

Share to your agent

Export one clean Markdown file. Download it, or paste it straight into your LLM or agent as ready-to-read context.

download .md
Online + offline

Wherever the meeting happens, Kasete is already there.

OFFLINE · IN THE ROOM

The meeting that's actually in the room

Standups, client visits, interviews, lectures. Set your phone on the table and let the mic capture every voice. Diarization keeps them apart.

MicrophoneIn-personInterviews
ONLINE · ON THE CALL

The call living inside a browser tab

Zoom, Meet, Teams, a webinar, a podcast. Share the tab with audio and Kasete records the call itself. No bot joins, no awkward invite.

Tab audioZoom · MeetWebinars
The app

A recorder that already did the hard part.

Tap record. When you tap Done, Kasete uploads, diarizes and transcribes in the background. You come back to a finished, speaker-separated transcript.

9:41● ● ●
Record
12:47
MicrophoneTab audio
Recording… tap to pause
CancelDone
Recent3
Weekly sync — Growth
42:18 · completed
Customer call · Acme
28:05 · completed
1:1 with Dewi
16:40 · processing
01 · Record
9:41● ● ●
Weekly sync — Growth
Jun 3, 2026 · 09:30
Completed
diarized · 42:18
EN · ID
1
Speaker 1en00:00
Okay, let's kick off. Did everyone get the deck I sent over last night?
2
Speaker 2en00:07
Got it. The retention numbers on slide four are the headline for me.
3
Speaker 3id00:15
Setuju. Tapi kita perlu data churn dari kuartal lalu sebelum commit.
1
Speaker 1en00:24
Fair. I'll pull churn by cohort and drop it in before Thursday.
02 · Diarized transcript
Share to your agent

The transcript is just Markdown. Your agent speaks Markdown.

No bespoke integration, no API to wire up. Download the .md or copy it, then drop it into Claude, ChatGPT, or any agent that reads text.

weekly-sync.md
# Weekly sync: Growth
2026-06-03 · 42:18 · EN, ID
## Speaker 1
Okay, let's kick off. Did everyone get the deck?
## Speaker 2
The retention numbers on slide four are the headline.
## Speaker 3
Setuju. Tapi kita perlu data churn kuartal lalu.
paste
or attach
weekly-sync.md
Summarise the sync and pull out every action item with an owner.
Here are the action items from the Growth sync:
Speaker 1: Pull churn by cohort before Thursday
Speaker 2: Own the retention story on slide 4
Speaker 3: Send Q1 churn data to the team
Features

Everything the recording needs to become useful.

Speaker diarization

Knows who said what. Colour-coded speaker labels, with voiceprints that can persist across recordings.

Multi-language

Detects and tags language per segment. Switch between English and Indonesian mid-conversation.

Up to 2 hours

Long-form ready. Audio is chunked and uploaded as you talk, so a two-hour board meeting just works.

Clean Markdown export

Every transcript is a single .md file: versionable, greppable, paste-able into any tool or model.

Background processing

Tap Done and walk away. Recordings transcribe in the background and surface the moment they're ready.

Private by default

Row-level security scopes every recording to you. Audio lives in private storage. Nothing is public.

FAQ

Questions, answered.

Yes. Choose "Tab audio" and share the browser tab the call is running in with audio enabled. Kasete records the call directly. No bot joins the meeting.

Up to two hours. Audio is split into chunks and uploaded while you record, then transcribed in parallel, so length isn't a problem.

Transcription auto-detects language and tags it per segment. English and Indonesian are tested, and a single recording can mix both.

A speaker-separated transcript inside the app, plus a single Markdown (.md) file you can download. Title, timestamps, speakers and language tags included.

There's nothing to integrate. The export is plain structured Markdown, so you copy it or attach the .md to any LLM or agent and it reads as clean context.

Recordings live in private storage scoped to your account with row-level security. Nothing is public, and only you can read your files.