Multilingual PSOLA text-to-speech system

Enhancing VoIP service for ubiquitous communication in a campus WLAN with partial coverage

“…Changing the playout speed exploits human's insensitivity to minor modulations in the speed of a speech signal [28]. Many algorithms have been proposed in the literature for performing efficient and high quality time-scale modification of speech [30][31][32]. In this paper, we choose the WSOLA (Waveform Similarity Overlap-Add) algorithm [32] for time-scale modification of speech since it preserves the pitch period and requires only time domain operations of the speech without any frequency domain transformation.…”

A hidden Markov-model-based trainable speech synthesizer

“…As digital storage problems eased, and concatenative systems were more thoroughly researched, augmented diphone systems were introduced (Olive, 1990;Bigorgne et al, 1993;Lernout & Hauspie, 1999). In these systems, longer polyphone units were introduced to improve concatenation smoothness in those contexts in which diphones joined least smoothly.…”

Synthesis of Arabic from short sound clusters

“…units derived from a given phonetic context, and having fixed lengths. Typical units used in the past are diphones, triphones, and demisyllables, for example the polyphone approach (Bigorne et al, 1991) used on the multilingual PSOLA, and the demisyllabic approach (Kraft & Andrew, 1992;Portele, Sendlmeier & Hess, 1990) used to synthesize German.…”