A Dutch Dysarthric Speech Database for Individualized Speech Therapy Research

A Dutch Dysarthric Speech Database for Individualized Speech Therapy Research Emre Yılmaz1, Mario Ganzeboom1, Lilian Beijer2,3, Catia Cucchiarini1 and Helmer Strik1,4 1Centre for Language and Speech Technology (CLST), Radboud University, Nijmegen, Netherlands 2Sint Maartenskliniek, Nijmegen, Netherlands 3HAN University of Applied Sciences, Faculty of Health and Social Studies, Nijmegen, Netherlands 4Centre for Language Studies (CLS), Radboud University, Nijmegen, Netherlands {e.yilmaz, m.ganzeboom, c.cucchiarini, h.strik}@let.ru.nl, l.beijer@maartenskliniek.nl Abstract We present a new Dutch dysarthric speech database containing utterances of neurological patients with Parkinsons disease, traumatic brain injury and cerebrovascular accident. The speech content is phonetically and linguistically diversified by using numerous structured sentence and word lists. Containing more than 6 hours of mildly to moderately dysarthric speech, this database can be used for research on dysarthria and for developing and testing speech-to-text systems designed for medical applications. Current activities aimed at extending this database are also discussed. Keywords: Speech database, dysarthric speech, Parkinsons disease, automatic speech recognition, speech therapy 1. Introduction Dysarthria is a speech disorder caused by difficulties in neuromuscular control (Duffy, 1995) that lead to decreased speech intelligibility and communication impairment (Kent and Kim, 2003). Consequently, the life quality of dysarthric patients is negatively affected (Walshe and Miller, 2011) and they run the risk of losing contact with friends and rel- atives and eventually becoming isolated from the society. Research has shown that intensive therapy can be effective in (speech) motor rehabilitation (Ramig et al., 2001; Bho- gal et al., 2003; Kwakkel, 2006; Rijntjes et al., 2009), but various factors conspire to make intensive therapy expen- sive and difficult to obtain. Recent developments show that therapy can be provided without resorting to frequent face-to-face sessions with therapists by employing computer-assisted speech training systems (Beijer and Rietveld, 2011). According to the out- comes of the efficacy tests presented in (Beijer et al., 2014), the user satisfaction appears to be quite high. However, most of these systems are not yet capable of automatically detecting problems at the level of individual speech sounds, which are known to have an impact on speech intelligibility (De Bodt et al., 2002; Yunusova et al., 2005; Van Nuffelen et al., 2009). One of the problems in developing more advanced and effective automatic speech therapy systems is the lack of sufficient language resources to develop and test the algo- rithms. Motivated by this problem, this paper presents a new Dutch dysarthric speech database with recordings of word and sentence sets uttered by 16 speakers with mild to moderate dysarthria due to Parkinsons disease (PD), traumatic brain injuries (TBI) and cerebrovascular accident (CVA). These word and sentence sets are carefully designed to ensure phonetically rich and balanced utterances. The audio files are annotated with the orthographic transcrip- tions and detailed speaker information such as age, gender, speech intelligibility level and origin of dysarthria. This database combines the output of two independent data collection efforts for Dutch as spoken in the Nether- lands. The first dysarthric speech data collection of Dutch- speaking patients aimed to use the data for a pilot study …

Content truncated to 500 words