Speech recognition and understanding using cloud resources and soft computing in RFID deployment services and online playing games

在RFID部署服务和在线游戏中使用云资源和软计算进行语音识别和理解

• 批准号: 283139-2012
• 负责人: Selouani, SidAhmed
• 金额: $ 1.02万
• 依托单位: Université de Moncton
• 依托单位国家: 加拿大
• 项目类别: Discovery Grants Program - Individual
• 财政年份: 2012
• 资助国家: 加拿大
• 起止时间: 2012-01-01 至 2013-12-31
• 项目状态: 已结题
• 来源: https://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=507909
• 关键词: Speech; recognition; understanding; using; cloud

Speech-enabled interfaces are especially useful in mobile devices due to the limited display and keyboard sizes. Despite the steady growth in the adoption of mobile devices in people's daily lives, and the prodigious progress of speech and language technologies, current systems are still a great distance away from providing satisfaction to users. In this program, new research directions are investigated in order to face this challenge. At low-level processing, cooperative approaches combining conventional and soft computing methods are proposed in various speech enhancement schemes to give speech recognition the ability to cope with adverse conditions. Soft computing is inspired by biological systems that are capable of dealing with uncertainty and imprecision to achieve robustness, tractability and optimal solutions. At high-level processing, current research still faces the challenge of giving dialogue systems the ability to perfectly understand the user intents within a few dialogue turns. The strategy consists of leveraging knowledge in distributed ontologies to reduce confusability and ambiguity and to provide a reliable semantic confidence measure through a new concept of "ontological distance". Both low-level and high-level processing are integrated in realistic applications by taking advantage of cloud computing and by providing versions for mobile operating systems. The first field of applications consists of allowing human operators to communicate verbally with Radio Frequency IDentification (RFID) devices and information systems. Hands-free speech systems combined with RFID technology are expected to improve the speed and safety of numerous business operations such as monitoring assets and shipping. The second field of applications will focus on online role playing games (RPGs), where an actor is mainly constructed through interactions with non-playing characters (NPCs). Our application will provide the option of interacting verbally with NPCs. Adding this feature will upgrade the gaming experience to a new level of intensity. The new frameworks will also be integrated in speech-enabled interfaces with the aim of enhancing the communication of both French and English-speaking users suffering from various speech disorders.

由于显示器和键盘尺寸有限，支持语音的界面在移动设备中特别有用。尽管移动设备在人们日常生活中的应用稳步增长，并且语音和语言技术取得了巨大进步，但当前的系统距离为用户提供满意的服务还有很长的路要走。在该计划中，研究新的研究方向以应对这一挑战。在低级处理方面，在各种语音增强方案中提出了结合传统计算方法和软计算方法的协作方法，以使语音识别能够应对不利条件。软计算受到生物系统的启发，生物系统能够处理不确定性和不精确性，以实现鲁棒性、易处理性和最佳解决方案。在高级处理方面，当前的研究仍然面临着让对话系统能够在几个对话回合内完美理解用户意图的能力的挑战。该策略包括利用分布式本体中的知识来减少混淆性和歧义性，并通过“本体距离”的新概念提供可靠的语义置信度度量。通过利用云计算和提供移动操作系统版本，低级和高级处理都集成在实际应用中。第一个应用领域包括允许操作员与射频识别 (RFID) 设备和信息系统进行口头通信。免提语音系统与 RFID 技术相结合有望提高资产监控和运输等众多业务运营的速度和安全性。第二个应用领域将集中于在线角色扮演游戏（RPG），其中演员主要通过与非玩家角色（NPC）的互动来构建。我们的应用程序将提供与 NPC 进行口头交互的选项。添加此功能将把游戏体验提升到一个新的强度水平。新框架还将集成到支持语音的界面中，旨在增强患有各种言语障碍的法语和英语用户的沟通。