Duration synthesis modelling

In the United States, compound RGW-2938 (47) has been developed as a selective positive inotropic agent its synthesis and pharmacological actions have been reviewed [161]. Related compounds, in which the dihydro-pyridazinone system is replaced by a benzoxazinyl system instead of the quinazoline system, have recently been studied in India and in the United States [162,163]. From this series, bemoradan (48) was found to be a very potent orally active, long-acting inotropic vasodilator agent in canine models (i.v. ED5o = 5.4 //g/kg 24h duration of action) [163]. 

Analysis/Synthesis With estimates of excitation and system sine-wave amplitudes and phases at the center of the new time-scaled synthesis frame, the synthesis procedure becomes identical to that of the baseline system of section 3.6. The goal then is to obtain estimates of the amplitudes, Ak (/), and phases, Ok (/), in Equation (9.42) at the center of the synthesis frame of duration Q = pQ where Q is the analysis frame interval as defined in section 3.5. Since in the time-scale modification model, the system and excitation amplitudes are simply time scaled, from Equations (9.40a) and (9.42) the composite amplitude need not be separated and therefore the required amplitude can be obtained from the sine-wave amplitudes measured on each frame m, by spectral peak-picking. 

The quantitative extent of CYP induction depends on the dosage (concentration) of the inducer and on the duration of exposure. However, the induction process, in contrast to inhibition, is not as straightforward to study in vitro, since induction requires intact cellular protein synthesis mechanisms as available in cell culture models (57-62). 

This disparity has been known for some time, and several attempts were made in ASR to correct this, the most notable proposal being the hidden semi-Markov model (HSMM) [282]. In this model, the transition probabilities are replaced by an explicit Gaussian duration model. It is now known that this increase in durational accuracy does not in fact improve speech recognition to any significant degree, most probably because duration itself is not a significant factor in discrimination. In synthesis, however, modelling the duration accurately is known to be important and for this reason there has been renewed interest in hidden semi-Markov models [504], [514],... 

HMM synthesis started with the now classic paper by Tokuda et al [453] which explained the basic principles of generating observations which obey the dynamic constraints. Papers explaining the basic principle include [452], [454] [305]. From this, a gradual set of improvements have been proposed, resulting in today s high quality synthesis systems. Enhancements include more powerful observation modelling [509] duration modelling in HMMs [504], [515] trended HMMs [134], [133], trajectory HMMs [515], HMM studies on emotion and voice transformation [232], [429], [404], [505]. 

Krishna, N. S., Talukdar, P. P., Bali, K., and Ramakrishnam, A. G. Duration modeling for Hindi text-to speech synthesis system. In Proceedings of the International Conference on Speech and Language Processing 2004 (2004). 

Yoshimura, T., Tokuda, K., Masuko, T., Kobayashi, T., and Kitamura, T. Simultaneous modelling of spectrum, pitch and duration in hmm-based speech synthesis. In Proceedings of Eurospeech 1999 (1999). 

If we were to use HMMs for synthesis, we know that the specification gives us the required phone sequence and possibly a diuation for each phone. The phone sequence tells us which models to use in which order, but not which states to use, or which observations to generate from the state Gaussians. If we have a duration in the specification, this tells us how many observations we should generate from this model, but again not which states to generate from. The question then is, if we want to use the models for synthesis, how do we actually go about this ... 

---