Title :
Speaker-independent detection of child-directed speech
Author :
Schuster, Sebastian ; Pancoast, Stephanie ; Ganjoo, Milind ; Frank, Michael C. ; Jurafsky, Dan
Author_Institution :
Dept. of Comput. Sci., Stanford Univ., Stanford, CA, USA
Abstract :
Identifying the distinct register that adults use when speaking to children is an important task for child development research. We present a fully automatic, speaker-independent system that detects child-directed speech. The two-stage system uses diarization-style voice activation techniques to extract speech segments followed by a supervised ν-SVM classifier trained on 1582 prosodic and log Mel energy features. The system significantly improves the state of the art, detecting child-directed speech with F1 of .66 (exact boundary) and .83 (within 1 second). A feature analysis confirms the importance of F0 features (especially 3rd quartile and range) as well as new features like the variance, kurtosis, and min of log Mel energy within a frequency band.
Keywords :
learning (artificial intelligence); signal classification; speech recognition; support vector machines; F0 features; F1 features; automatic speaker-independent system; child development research; child-directed speech detection; child-directed speech improvement; diarization-style voice activation techniques; exact boundary; feature analysis; frequency band; kurtosis feature; log Mel energy features; prosodic features; speech segment extraction; supervised ν-SVM classifier training; two-stage system; variance feature; Accuracy; Gold; Measurement; Noise; Speech; Support vector machines; Training; Child-directed Speech; Language Development; Prosody; Speech Analysis;
Conference_Titel :
Spoken Language Technology Workshop (SLT), 2014 IEEE
DOI :
10.1109/SLT.2014.7078602
