Backed by 10+ years of academic research, our technology detects, segments, counts, and labels speakers, making transcription and analysis faster and more efficient.
Access our beta program to discover our cutting edge speaker diarization models
STATE-OF-THE-ART
Discover the power
of speaker diarization
Our AI speaker diarization models accurately identify and separate speakers in audio recordings, providing valuable insights and improving productivity.
USE CASES
Explore the ways in which pyannote users incorporate our technologies into their tech stack to deliver top-tier products
OPTIMIZED BY DESIGN
Our optimized AI models accurately separates and identifies speakers in audio recordings, saving you time and effort.
x2 faster*
+ 20% accuracy*
*versus pyannote open source model
ENHANCED FEATURES
Speaker diarization
Partition multi-speaker conversations into separate speakers
Speaker identification
Track specific speakers across multiple conversations using voiceprints
Overlapping speech
Flag when multiple speakers talk over each other
Change point detection
Mark speaker change points
Voice activity detection
Spot when anyone is speaking
Speaker separation
Isolate speech of overlapping speakers
Confidence score
Pinpoint the exact areas where human attention is required