Most voice note apps stop at the transcript.
You record a thought, hit stop, and the app gives you a paragraph of text. Then what. The transcript sits in a folder you forget about. Search finds keywords, not ideas. There is no link between this recording and the client it is about, the project it belongs to, or the meeting where the topic first came up. You end up with a graveyard of voice memos and a brain that still does not know what you said. That is not a thinking partner. That is an archive.
Speaker labels
Diarisation separates who said what. Names attach the same way they do for any meeting transcript.
100+ languages, picked automatically
Language detection runs on every recording, so your Greek voice memo and your English client call both come out right.
Up to 150 MB per file
Roughly an hour and a half at typical voice-memo bitrate. Drop a long walk, a workshop, a workshop debrief.
How your digital brain handles a voice note
- 1
Record or upload
Hit record in the browser, or upload an audio file from your phone, your laptop, or a Zapier flow watching your Krisp or tl;dv recordings. Up to 150 MB per file. Common formats supported (mp3, m4a, wav, ogg, webm, aac, flac, opus, mp4).
- 2
Transcribe with speakers, in any language
Your brain transcribes with AssemblyAI. Speaker diarization labels who said what without you tagging anything. Language detection picks the right model automatically, so a Greek voice memo and an English client call both come out right.
- 3
Connect to everything else you know
The transcript runs the same pipeline as a meeting transcript or a document. People, projects, decisions, dates, and topics get extracted and linked to entities already in your brain. Tomorrow you ask 'what did the founder of Acme say about pricing last quarter,' and the answer cites the voice note you recorded on the train.
What you walk away with
- Voice memos that turn into searchable knowledge, not transcript debris
- Speaker labels and language detection out of the box (100+ languages)
- Voice notes sit beside documents, URLs, and YouTube in the same brain
- Trigger it from Claude ("save this voice memo") or Zapier in the background
- You stay in the loop. The brain captures, you decide what matters
Trigger it from where you already work
Frequently asked questions
What audio formats can I upload?+
mp3, m4a, wav, ogg, oga, opus, webm, aac, flac, and mp4 audio. Most phone-recorder defaults are covered. The size cap is around 150 MB per file, which is roughly an hour and a half at typical voice-memo bitrates.
Which languages are supported?+
Over 100 languages. Language detection is automatic. You do not pick the language, your brain hears the recording and runs the right model. Mixed-language recordings get the dominant language.
Will it identify who is speaking?+
Yes. Speaker diarization labels each utterance (Speaker A, Speaker B…) so a two-person voice memo or a small group recording stays readable. Names attach later, the same way they do for meeting transcripts in your brain.
Can I record in the browser, or only upload?+
Both. The record button is built into the upload flow. Hit record, talk, hit stop, it goes straight in. For longer or higher-quality recordings, upload the file your phone or recorder already made.
How is this different from Otter or Fireflies?+
Otter and Fireflies are meeting bots. They join calls and produce transcripts you read later. BrainGraph is the layer underneath: voice notes (and meeting transcripts, documents, URLs, YouTube) land in one digital brain that connects them to people, projects, and decisions. You can use Otter or Fireflies for the calls and BrainGraph for the brain that remembers them.
Can I trigger this from Claude or Zapier?+
Yes to both. In Claude, the MCP connector exposes an upload_audio_url tool: say 'save this voice note: <link>' and Claude calls it. In Zapier, point a Zap at your brain and any new Krisp, tl;dv, or Google Drive audio flows in automatically.
Is my audio private?+
Yes. Uploaded audio is processed for transcription and stays in your private brain. Content is never used to train AI models. See the privacy policy for the full details.
What does a voice note count as against my plan?+
One audio file is one document against your plan limit, same as a PDF or a meeting transcript. A failed transcription refunds the slot, so you do not lose a document to a glitch.
Stop transcribing into a graveyard.
Voice notes that connect to the rest of what you know. Start free with 40 documents.