This article proposes a new technique for skimming multimedia content such as video mail, audio/visual data in blog sites, and other consumer-generated media. The proposed method, which is based on the automatic extraction of emphasized speech, locates emphasized portions of speech with high accuracy by using prosodic parameters such as pitch, power, and speaking rate. As the method does not employ any speech recognition technique, it enables a highly robust estimation in noisy environments. To extract emphasized portions of speech, the method introduces a metric, “degree of emphasis,” which indicates the degree of emphasis of each speech segment. Given an article, the method computes the degree of emphasis for each speech segment in it. When a user requests a skimming of the article's content, the method refers to the user-specified “skimming rate” to collect the emphasized segments. Preference experiments were performed in which participants were asked to select either the skimmed contents created by our method or those created using a fixed interval approach. The preference rate of our method was about 80%, which suggests that the proposed method can generate proper content skimming.

This article describes a basic study on design guidelines for a public work environment where personal information might be handled. In this study, we conducted a user survey to extract factors that affect reassurance and performed an experiment to evaluate the effect of partition size on reassurance. First, from the user survey, we extracted personal-information and work-environment factors that affect reassurance in a public work environment. Next, we examined the effect of partition size (height and depth) on reassurance and found that “the ability of someone viewing my monitor from a neighboring seat” is a factor affecting reassurance and that “the ability of seeing the figure (feel the presence) of a neighbor” does not affect reassurance. In short, partition height has the role of blocking information (on the monitor) from a neighbor, and provided that partition height satisfies a certain value (keeping the monitor out of a neighbor's view), partition depth will also make a significant difference in reassurance. The role of partition depth, moreover, is not so much to block information but rather to block off the user's space.

We propose Choco-Para, a multimedia content skimming technique; its application to a variety of content types is described. Based on automatic speech emphasis extraction, Choco-Para extracts speech attributes, prosodic parameters such as pitch, power, and speaking rate, and uses the data to estimate the degree of emphasis of each spoken phrase. By computing the degree of the emphasis curve, Choco-Para can generate a skimmed edition at an arbitrary skimming rate by selecting emphasized speech portions via dynamic threshold logic. Choco-Para uses three types of prosodic parameters and both short term and long term deviation. Experiments assess the contributions of each prosodic parameter and deviation type. They show that estimation accuracy is optimized by using both short and long term deviation with regard to pitch, power, and speaking rate. The results confirm that Choco-Para supports a wide variety of multimedia content

This paper proposes a new text-to-speech (TTS) system that utilizes large numbers of speech segments to produce very natural and intelligible synthetic speech. There are two innovations; new multiform synthesis units and a new speech modification algorithm based on a vocoder that offers harmonics reconstruction. The multiform units make it possible to reduce acoustic discontinuities at concatenation points and unnatural sound by preparing synthesis units with various lengths and various F/sub 0/ contours. The new speech modification algorithm, on the other hand, improves the quality of prosody modified speech. This algorithm is extremely effective in synthesizing speech whose prosodic parameters are quite different from those of synthesis units. Listening tests confirm that the new synthesis units yield speech with high intelligibility and naturalness, and that the new speech modification algorithm is superior to all other conventional vocoders and waveform domain algorithms including TD-PSOLA, especially when modifying the F/sub 0/ frequency upward.

In this paper, we first introduce the use of Gaussian mixture models (GMM) for Chinese tone classification in continuous speech. Then, we explain how to integrate it with the HMM-based speech recognition system. Finally, we provide the tone classification accuracy of this probabilistic method which is tested with Chinese continuous speech database of national “863” project.

The Multi-layered Speech/Sound Synthesis Control Language (MSCL) proposed herein facilitates the syn-thesizing of several speech modes such as nuance, mental state and emotion, and allows speech to be synchronized to other media easily. MSCL is a multi-layered linguis-tic system and encompasses three layers: and seman-tic level layer (The S-layer), interpretation level layer (The I-layer), and parameter level layer (The P-layer). The S-layer is the description level of semantics such as emotional and emphasized speech. The I-layer is the description level of prosodic feature controls and inter-prets The S-layer scripts to for control on I-layer level. The P-layer represents prosodic parameters for speech synthesis. This multi-level description system is conve-nient for both laymen and professional users. MSCL also encompasses many eective prosodic feature con-trol functions such as a time-varying pattern descrip-tion function, absolute and relative control forms, and SDS(Speaker Dependent Scale). MSCL enables more emotional and expressive synthetic speech than conven-tional TTS systems. This paper describes these func-tions and the eective prosodic feature controls possible with MSCL. 1

This paper proposes a high quality speech synthesis system based on waveform synthesis. Our most recent text-to-speech (TTS) system achieved very good performance by using waveform synthesis units. The system, however, required a very large speech data base. A compact system could only be created by developing a very small data package. This paper proposes new approach to speech synthesis based on waveform segments. A novel point of the approach is its simple algorithm which effectively forms synthesis unit package using the clustering technique. The algorithm is applied to our TTS system. Evaluation by listening tests confirm the good performance of the algorithm.>

The authors propose a text-to-speech synthesis method based on automatic synthesis unit generation techniques using a natural speech database. They have termed the automatic procedure context oriented clustering (COC). Using the COC procedure, 627 phonetic synthesis units were generated automatically based on 432 words uttered by a male speaker. This systematic approach has several advantages. First, as synthesis units can be generated automatically without any a priori phonological knowledge, it is easy to change the number of units and voices. Second, following from this, the technique can be applied to any language. Third, the generation of allophonic synthesis units is not dependent on the human decisions but on the statistical characteristics of spectral parameters in natural speech. Thus, the generated units are more consistent that those obtained through other methods, with the result that more intelligible speech can be reconstructed.>