You are here
Speaker identification aims to decide who is the speaking person among a set of candidate individuals or to establish that the speaker is unknown.
Speaker verification verifies a person’s claimed identity from her/his voice.
In both cases, the task becomes very difficult when the subject speaks at a distance of some meters from the microphones, under noisy and reverberant conditions. The task also regards the processing of audio streams, often combined with speech activity detection to determine the task of speaker diarization.
Related keywords: robust acoustic features for speaker recognition, speaker modeling and characterization, speaker confidence estimation, pitch and formant estimation, robustness to variable and degraded channels, multimodal and multimedia human speaker recognition.