SPPAS - MULTI-LINGUAL APPROACHES TO THE AUTOMATIC ANNOTATION OF SPEECH

The first step of most acoustic analyses unavoidably involves the alignment  of recorded speech sounds with their phonetic annotation. This  step is very labor-intensive and cost-ineffective since it has to be performed manually by experienced phoneticians during many hours of work. This paper describes the main features of SPPAS, a software tool designed for the needs of automatically producing annotations of speech at the level of utterance, word, syllable and phoneme based on the recorded speech sound  and  its  orthographic transcription. In other words, it can automatize the phonetic transcription task for speech materials, as well as the alignment task of transcription with speech recordings for further acoustic analyses. 
Special attention will be given to the methodology implemented in SPPAS, based on algorithms   which are as language-and-task-independent as possible. This  procedure allows for the addition of new languages quickly and for the adaptation of this tool  to  the  user's specific needs. Consequently,  the  quality  of  the  automatic annotations is largely influenced by external resources, and the users can modify the process as needed. In that sense, phoneticians need automatic tools and these tools can be significantly improved by phonetician input.