In its nine chapters, this book provides an overview of the state-of-the-art and best practice in several sub-fields of evaluation of text and speech systems and components. The evaluation aspects covered include speech and speaker recognition, speech synthesis, animated talking agents, part-of-speech tagging, parsing, and natural language software like machine translation, information retrieval, question answering, spoken dialogue systems, data resources, and annotation schemes. With its broad coverage and original contributions this book is unique in the field of evaluation of speech and language technology. This book is of particular relevance to advanced undergraduate students, PhD students, academic and industrial researchers, and practitioners.
For a machine to convert text into sounds that humans can understand as speech requires an enormous range of components, from abstract analysis of discourse structure to synthesis and modulation of the acoustic output. Work in the field is thus inherently interdisciplinary, involving linguistics, computer science, acoustics, and psychology. This collection of articles by leading researchers in each of the fields involved in text-to-speech synthesis provides a picture of recent work in laboratories throughout the world and of the problems and challenges that remain. By providing samples of synthesized speech as well as video demonstrations for several of the synthesizers discussed, the book will also allow the reader to judge what all the work adds up to -- that is, how good is the synthetic speech we can now produce? Topics covered include: Signal processing and source modeling Linguistic analysis Articulatory synthesis and visual speech Concatenative synthesis and automated segmentation Prosodic analysis of natural speech Synthesis of prosody Evaluation and perception Systems and applications.
This book constitutes the refereed proceedings of the Second International Workshop on Text, Speech and Dialogue, TSD'99, held in Plzen, Czech Republic in September 1999. The 57 revised full papers and 19 posters presented were carefully reviewed and selected for inclusion in the book. The book presents state-of-the-art research and technology in the field of natural language processing with emphasis on text, speech, and spoken dialogue.
The quality of a telecommunication voice service is largely inftuenced by the quality of the transmission system. Nevertheless, the analysis, synthesis and prediction of quality should take into account its multidimensional aspects. Quality can be regarded as a point where the perceived characteristics and the desired or expected ones meet. A schematic is presented which classifies different entities which contribute to the quality of a service, taking into account conversational, user as weIl as service related contributions. Starting from this concept, perceptively relevant constituents of speech communication quality are identified. The perceptive factors result from ele ments of the transmission configuration. A simulation model is developed and implemented which allows the most relevant parameters of traditional trans mission configurations to be manipulated, in real time and for the conversation situation. Inputs into the simulation are instrumentally measurable quality elements commonly used in transmission planning of telephone networks. A reduced set of these quality elements forms a basis for models which aim at predicting mouth-to-ear quality as it would be perceived by a user of the sys tem. These models are an important tool for the planner of telecommunication networks, as they allow the expected quality to be estimated in advance, even before the network has been set up. Two well-known models (the SUBMOD and the E-model) are analyzed in more detail, with an emphasis on the psy choacoustic and psychophysical backgrounds.
Dictation systems, read-aloud software for the blind, speech control of machinery, geographical information systems with speech input and output, and educational software with `talking head' artificial tutorial agents are already on the market. The field is expanding rapidly, and new methods and applications emerge almost daily. But good sources of systematic information have not kept pace with the body of information needed for development and evaluation of these systems. Much of this information is widely scattered through speech and acoustic engineering, linguistics, phonetics, and experimental psychology. The Handbook of Multimodal and Spoken Dialogue Systems presents current and developing best practice in resource creation for speech input/output software and hardware. This volume brings experts in these fields together to give detailed `how to' information and recommendations on planning spoken dialogue systems, designing and evaluating audiovisual and multimodal systems, and evaluating consumer off-the-shelf products. In addition to standard terminology in the field, the following topics are covered in depth: How to collect high quality data for designing, training, and evaluating multimodal and speech dialogue systems; How to evaluate real-life computer systems with speech input and output; How to describe and model human-computer dialogue precisely and in depth. Also included: The first systematic medium-scale compendium of terminology with definitions. This handbook has been especially designed for the needs of development engineers, decision-makers, researchers, and advanced level students in the fields of speech technology, multimodal interfaces, multimedia, computational linguistics, and phonetics.
Speech processing addresses various scientific and technological areas. It includes speech analysis and variable rate coding, in order to store or transmit speech. It also covers speech synthesis, especially from text, speech recognition, including speaker and language identification, and spoken language understanding. This book covers the following topics: how to realize speech production and perception systems, how to synthesize and understand speech using state-of-the-art methods in signal processing, pattern recognition, stochastic modelling computational linguistics and human factor studies.
Languages, in all their forms, are the more efficient and natural means for people to communicate. Enormous quantities of information are produced, distributed and consumed using languages. Human language technology's main purpose is to allow the use of automatic systems and tools to assist humans in producing and accessing information, to improve communication between humans, and to assist humans in communicating with machines. This book, sponsored by the Directorate General XIII of the European Union and the Information Science and Engineering Directorate of the National Science Foundation, USA, offers the first comprehensive overview of the human language technology field.
