The Role of Speech Science in Developing Robust Speech Technology Applications

本次研讨会将集中讨论语音科学在开发稳健的未来语音技术应用中应发挥的关键作用。研讨会的结果是一份报告，其中载有向国家科学基金会提出的动机良好的跨学科核心建议，内容是如何最好地促进和促进语音科学在开发语音技术应用方面的作用，以造福社会。特别令人感兴趣的是那些行业不太可能投资的领域，因为主题不是优先事项，时间进程太长，或者服务的人群不感兴趣。研讨会的次要影响还可能包括与会者对促进当地大学或机构的言语科学研究感兴趣的具体未来合作。研讨会将成为一个独特的机会，通过聚集三个小组：(1)言语科学家，(2)言语工程师，和(3)政府项目经理，跨越多个主题领域，讨论言语科学和言语工程的跨学科视角。受邀者将包括科学家和技术专家，他们是根据NSF项目经理和该领域专家的推荐而选择的。研讨会受邀者将被要求对具体选定的研究领域进行简要概述，解释他们认为需要取得更多进展的领域，并就取得进展的障碍提出意见。主题领域将包括大型语音公司(包括儿童和老年人)往往不是主要关注点的人群的语音处理，以及尚未很好理解的问题，如说话人的可变性、上下文影响和说话人状态、生物模型以及从小样本集学习。来自这些不同学科和主题领域的与会者通常不会聚在一起讨论未来的研究方向；受邀者名单的制定正好考虑到了这一目标。

This workshop will focus on the critical role that speech science should play in developing robust future speech technology applications. The result of the workshop is a report containing well-motivated interdisciplinary core recommendations to NSF on how to best promote and foster the role of speech science in developing speech technology applications for the long-term benefit of society. Of special interest are those areas in which industry is unlikely to invest in because the topic is not a priority, the time course is too long, or the population served is not of interest. Secondary impacts of the workshop may also include specific future collaborations among attendees interested in promoting speech science research at their local university or institute.The workshop will serve as a unique opportunity for discussing an interdisciplinary perspective on speech science and speech engineering by bringing together three groups: (1) speech scientists, (2) speech engineers, and (3) government program managers, across multiple topics areas. Invitees will include scientists and technologists, chosen based on recommendations from NSF Program Managers and from experts in the field. The workshop invitees will be asked to give a brief overview of specifically chosen research areas, explain where they think more progress is needed, and give opinions on the barriers to progress. Topic areas will include speech processing for populations often not of prime focus for large speech companies, including children and the elderly, as well as problems that are not well understood such as speaker variability, contextual affects and speaker state, biological models, and learning from small sample sets. Participants from these different disciplines and topic areas typically do not meet together to discuss future research directions; the list of invitees has been developed with exactly this goal in mind.