Assessing the effectiveness of diarization algorithms in costa rican children-adult speech according to age group and gender

Speaker diarization is the task of automatically identifying speaker identities and detecting their speaking times in an audio recording. Several algorithms have shown improvements performance this during past years. However, it still has challenges interaction scenarios, such as between a child adult, where interruptions, fillers, laughs other elements may affect detection clustering segments.
 In work, we perform exploratory study with two children-adult interactions within recording studio assess effectiveness different age groups genders. All participants are native Costa Rican Spanish speakers. The children ages 3 to 14 years, combines guided repetition words or short phrases, well natural speech.
 results demonstrate how affects performance, both cluster purity purity, direct but non-linear fashion.