Speaker diarization

Who said what, with names you actually edit

AudioMap detects each speaker, lets you name them once, and recognizes them across future recordings. Voiceprints are biometric data under GDPR Art. 9 and stay on our own servers in Germany.

Try for freeSee pricing
🗣️

Speaker detection in every paid tier

Not gated to Enterprise like other tools. From Personal €9 you get diarization on every transcript.

✍️

Editable names

Replace "Speaker 1" with "María" once: AudioMap remembers across future recordings. No re-labeling every meeting.

🇪🇺

Voiceprints in Europe

A voiceprint is biometric data under GDPR Art. 9. Yours stay on our own Hetzner DE servers, so the legal basis and the residency are ours to answer for, not a US processor’s.

🧠

Fine-tuned for Spanish voices

Speaker boundaries trained on Iberian + LATAM voices, not generic English. Better separation in fast-paced Spanish conversation.

👥

Multi-speaker right of access

AEPD April 2026 compliance: right of access applies in multi-speaker recordings. We provide structured exports per speaker.

🔄

Re-label after editing

Misidentified speakers? Edit once, AudioMap retrains the fingerprint. No need to re-transcribe.

How accurate is diarization for Spanish?+

In our tests with multi-speaker Spanish conversations, AssemblyAI gets 92–96% speaker assignment accuracy. We don't use Whisper for diarization (Whisper doesn't support it). For Whisper-only mode (fallback), diarization is unavailable for that recording.

What happens to the voiceprint when I delete my account?+

It is deleted together with your account data when you delete the account, and you can purge your voice profiles at any time. Details at /legal/aepd-compliance.

Can someone request deletion of their voice if they're not a user?+

Yes. /legal/request-deletion endpoint is public: a third party who appears in your recording can request removal without registering with us. This is the AEPD April 2026 requirement other competitors do not implement.

Does it work with overlapping speech?+

Partially. We separate primary speakers reasonably well; cross-talk gets attributed to the dominant voice. For perfect overlap detection use a multi-mic setup.

You may also like

vs Ottervs Firefliesvs Fathomvs Granola
AI transcription
Spanish from Spain and Latin America + EU data residency
AEPD compliance
Spanish DPA April 2026: point by point

Know who said what in every meeting

Speaker diarization on every transcript from €9/mo. Voiceprints on our own servers.

Try for free