Automatic Segmentation and Labeling for Spontaneous Standard Malay Speech Recognition

In this paper, we proposed an automatically segmenting and transcribing spontaneous speech signal without the use of manually annotated speech database. The spontaneous speech signal is first segmented into syllable-like units by considering short-term energy as a magnitude spectrum of some arbitrary signal. Similar syllable segments are then grouped together using an unsupervised incremental clustering technique. Separate models are generated for each cluster of syllable segments. At this stage, labels are assigned for each group of syllable segments manually. The syllable models of these clusters are then used to transcribe or recognize the spontaneous speech signal of closed-set speakers' data as well open-set speaker data. As a syllable recognizer, our initial results on Standard Malay television (TV3) news bulletins of the native and non-native speakers shows that the performance is 42.53% and 30.8% respectively.