首页> 外文会议>コンピュータセキュリティシンポジウム2018論文集 >Towards Practical Secure Automatic Speech Recognition
【24h】

Towards Practical Secure Automatic Speech Recognition

机译:迈向实用安全的自动语音识别

获取原文
获取原文并翻译 | 示例

摘要

Smart speakers are become popular thanks to the improvements of automatic speech recog-nition systems. However, the always-on system raises privacy problems. Moreover, we imagine inthe future that the service provider may change the smart speakers to be active that the speakerswill offer user-interested/user-aimed information according to the user conversations without beingfirstly evoked, which brings the further privacy concerns. In this study, we address this privacy issueand discover the feasibility of realizing such active smart speakers. Specifically, we mainly investi-gated the computation cost of using the state-of-the-art multiplication triples generation protocols toprepare for three different speech recognition models and give detailed results. The results show thatthe DNN model is the most practical secure ASR model at present with existing protocols becauseit only needs 6.35 hours for preparing private evaluation in the remaining 17.65 hours of a day.
机译:智能扬声器由于自动语音识别系统的改进而变得流行。但是,永远在线系统会引起隐私问题。此外,我们可以想象,将来服务提供商可能会将智能扬声器更改为活动扬声器,从而使扬声器会根据用户对话提供用户感兴趣/用户瞄准的信息,而不会被首先引起,这带来了更多的隐私问题。在这项研究中,我们解决了这个隐私问题,并发现了实现此类有源智能扬声器的可行性。具体来说,我们主要研究了使用最新的乘法三元组生成协议来准备三种不同的语音识别模型的计算成本,并给出了详细的结果。结果表明,DNN模型是现有协议中最实用的安全ASR模型,因为在一天的剩余17.65小时中,仅需要6.35小时即可准备私人评估。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号