Clinical transcription
The clinical transcription is the foundation of all nexmin analysis: who speaks and when, word by word, with the non-verbal events that matter in consultation — laughter, crying, sighs, long silences — placed at their real moment. It auto-detects the session's language, keeps the client's quotes in their original language, and is hand-editable: your correction enters the record as if you had written it yourself.
What sets it apart from generic transcription is the terrain: consultation voice. Low, emotional speech, overlaps, silences that last, peninsular and Latin American dialects. On that terrain, nexmin delivers word-level timestamps — not just per turn — client/therapist diarization, and a voice-assignment step that proposes itself and the therapist confirms. Non-verbal events are treated as clinical data, not noise: a laugh while speaking of abandonment is not an audio artefact, it is a signal. They are inserted exactly where they occur, and voice analysis reads them alongside prosody. On languages: the system identifies whichever is spoken in the session — Catalan is transcribed literally, not translated — and the analysis comes out in the therapist's language while keeping the original verbatim of the quotes. And as with everything in nexmin, nothing is signed without you: the transcription is editable, and wherever the model mishears, your correction wins.
Inside nexmin
The first tab of every analysed session. The hours allowance is consumed here: engines that work on already-transcribed text do not subtract minutes.
lectura larga: la página completa sobre este tema →
Related terms
Last updated: 2026-07-28