Phonetic Feature Discovery in Speech using Snap-Drift
Sin Wee Lee, Dominic Palmer-Brown · UEL Research Repository (University of East London) · 2006
This paper presents a new application of the snapdrift algorithm [1]: feature discovery and clustering of speech waveforms from nonstammering and stammering speakers. The learning algorithm is an unsupervised version of snapdrift which employs the complementary concepts of fast, minimalist learning (snap) & slow drift (towards the input pattern) learning. The SnapDrift Neural Network (SDNN) is toggled between snap and drift modes on successive epochs. The speech waveforms are drawn from a phonetically annotated corpus, which facilitates phonetic interpretation of the classes of patterns discovered by the SDNN.