Automatic labeling of prosodic events in speech has potentially significant implications for spoken language processing applications, and has received much attention over the years, especially after the introduction of annotation standards such as ToBI. Current labeling techniques are based on supervised learning, relying on the availability of a corpus that is annotated with the prosodic label...

Northern Bizkaian Basque shares important prosodic features with Tokyo Japanese, including the existence of a lexical distinction between accented and unaccented content words, the presence of phrase-initial rises and the consistent realization of accents as tonal falls. In this paper we investigate whether NB Basque is also like Japanese in not making use of syllable duration as a correlate of...

One of the main aspects in text-to-speech synthesis is the successful prediction of prosodic events. In this work we deal with the prediction of prosodic phrase breaks, accent tones and boundary tones from a linguistically XML-based enriched input (SOLE-ML) produced by a Natural Language Generator (NLG) system. We first extended the original specification of SOLE-ML in order for the NLG to prod...

The linguistic features analysis for input text plays an important role in achieving natural prosodic control in text-to-speech (TTS) systems. In a conventional scheme, experts refine suspicious if-then rules and change the tree structure manually to obtain correct analysis results when input texts that have been analyzed incorrectly. However, altering the tree structure drastically is difficul...

A new method was developed to include prosodic boundary information into statistical language modeling. This method is based on counting word transitions separately for the cases crossing accent phrase boundaries and not crossing them. Since direct calculation of the above two types of word transitions requires a large speech corpus which is practically impossible to make, bi-gram counts of par...

A new set of context labels was developed for HMM-based speech synthesis of Japanese. The conventional labels include those directly related to sentence length, such as number of “mora” and order of breath group in a sentence. When reading a sentence, it is unlikely that we count its total length before utterance. Also a set of increased number of labels is required to handle sentences with var...

The topic of this paper is the variability of phrase-initial pitch patterns of South Swedish. Central Swedish pitch patterns for phrase-initial accent I both to the East (Stockholm) and to the West (Gothenburg) display an apparent constancy, albeit with distinct patterns: East Central Swedish rising and West Central Swedish fall-rise. In South Swedish, the corresponding pitch patterns can be de...

Japanese numbers display a bewildering variety of pitch-accent behavior, and textbooks that address this problem at all provide pitch patterns only for small numbers and for large round numbers. Most large numbers in actual conversation, of course, have to be assembled on the fly rather than retrieved from memory. Nonetheless, the pitch patterns that occur on large numbers are quite systematic....

A new statistical language modeling was proposed where word n-gram was counted separately for the cases crossing and not crossing accent phrase boundaries. Since such counting requires a large speech corpus, which hardly can be prepared, part-of-speech (POS) n-gram was first counted for a small-sized speech corpus for the two cases instead, and then the result is applied to word n-gram counts o...

In this paper a method for the automatic labeling of phrase accents is described, based on a large text corpus that has been generated automatically and read by 100 speakers. Perception experiments on a subsetof 500 utterances show a high agreement between the automatically generated accent labels and the judgment scores obtained. We computed different prosodic feature vectors from the speech s...