This book presents guidelines for manual segmentation of the speech signal based on acoustic, articulatory, and perceptual features of speechsounds. It deals with transitions between various types of speechsounds pronounced both canonically and in a non-standard way, mostly exploiting visual information in the spectrogram and in the waveform. The objective is to provide for uniform segmentation of phonetic corpora based on phonetically motivated and easily applicable rules.
The book is designed for anyone working with human speech, whether it is phoneticians, speech technologists, or psycholinguists. That is why prior knowledge of only very elementary concepts is assumed, like “what does the spectrogram show” or “what is the formant”.
The book is out in Czech and English language.