A Portable Interactive Environment for Computer Speech Research

需要具有交互界面的强大的、可扩展的语音研究工具。这一努力是为了产生一套集成的软件工具来实现语音研究，并对研究基础设施做出贡献。将包括的工具支持语音分析、语音产生和语音识别，特别是从声学的角度。神经网络模拟工具提供了模拟具有生物学似然性的识别工具的能力。这些工具还可以用来教授学生有关语音和计算机语音识别的知识。具体目标是：(A)加强现有的软件研究工具；(B)开发一个可移植和可扩展的环境；(C)为用户编写文件；(D)分发软件。这些工具将免费提供，并将随附有文档的源代码一起分发，以便在本地进行增强。由于该工具包将可移植到多个平台上，因此它可以缩短在每个语音研究实验室中重新发明相同工具的需要，这些工具带有不同的错误和错误。学生将能够在毕业时使用在后续研究和教学中有用的工具包，后续研究的成本可能会降低。这些工具的潜在用户社区包括计算机语音识别研究人员、语音病理学家、语音心理学家、语言学家等。

There is a need for powerful, extensible speech research tools with an interactive interfaces. This effort is to produce an integrated set of software tools to enable speech research and is a contribution to research infrastructure. The tools to be included support speech analysis, speech production and speech recognition, particularly from the acoustic point of view. The neural network simulation tools provide the ability to simulate recognition tools with biological plausibility. The tools may also be used for teaching students about speech and computer speech recognition. The specific goals are to (a) enhance existing software research current tools, (b) develop an portable and extensible environment, (c) develop documentation for users, and (d) provide distribution of the software. The tools will be available free of charge and will be distributed with documented source code so they can be enhanced locally. Since the toolkit will be portable to several platforms, it can short-circuit the need to reinvent the same tools, with different errors and bugs, in each speech research laboratory. Students would be able to graduate with a toolkit useful in subsequent research and teaching and the cost of subsequent research may be reduced. The communities of potential users of these tools include computer speech recognition researchers, speech pathologists, speech psychologists, linguists, and others.