Profiles for people in your life, auto-updated from voice and chat data.
Loading...
Export a chat from WhatsApp (three dots β More β Export chat) and drop the .txt file below. Messages will be added to your searchable history.
When enabled, audio clips from unrecognized speakers will be saved so you can later assign them to identities.
Maximum number of clips to save for each unknown speaker.
Compensates for late music detection. If "remaining time" is consistently off, adjust this.
Positive (+) = detection was late, subtract from remaining time
Negative (-) = detection was early, add to remaining time
Controls how aggressively similar voices are grouped into "Person X" labels.
Lower (70-80%) = More aggressive merging (fewer Person labels)
Higher (85-95%) = More conservative (more Person labels, better differentiation)
Maximum character length for transcripts. Longer texts (likely from videos) are skipped.
Requires Mentra restart to take effect.