边缘的副语言隐私保护
Paralinguistic Privacy Protection at the Edge
摘要：
语音用户界面和数字助理正在迅速进入我们的生活，并成为跨越我们设备的单一接触点。这些始终在线的服务捕获我们的音频数据并将其传输到强大的云服务，以进行进一步处理和后续操作。通过这些设备收集的我们的语音和原始音频信号包含大量敏感的非语言信息，这些信息无论是故意的还是错误的触发都被传输给服务提供商。由于我们的情感模式和敏感属性，如我们的身份，性别，幸福，很容易使用深层声学模型推断，我们通过使用这些服务遇到了新一代的隐私风险。减轻基于语言的隐私泄露风险的一种方法是在传输语音数据之前利用基于云的处理与隐私保护、设备上的语言信息学习和过滤的组合。在本文中，我们介绍EDGY，一个可配置的，轻量级的，解开表示学习框架，转换和过滤高维语音数据，以识别和包含敏感属性的边缘卸载到云之前。我们评估EDGY的设备性能，并探索优化技术，包括模型量化和知识蒸馏，以实现资源受限设备上的私有，准确和高效的表示学习。我们的研究结果表明，EDGY运行在几十毫秒的“zero-shot”ABX分数或最小的性能损失约5.95%的字出错率（WER）在学习语言表示从原始语音信号，使用CPU和单核ARM处理器没有专门的硬件。
英文摘要：
Voice user interfaces and digital assistants are rapidly entering our lives and becoming singular touch points spanning our devices. These always-on services capture and transmit our audio data to powerful cloud services for further processing and subsequent actions. Our voices and raw audio signals collected through these devices contain a host of sensitive paralinguistic information that is transmitted to service providers regardless of deliberate or false triggers. As our emotional patterns and sensitive attributes like our identity, gender, well-being, are easily inferred using deep acoustic models, we encounter a new generation of privacy risks by using these services. One approach to mitigate the risk of paralinguistic-based privacy breaches is to exploit a combination of cloud-based processing with privacy-preserving, on-device paralinguistic information learning and filtering before transmitting voice data. In this paper we introduce EDGY, a configurable, lightweight, disentangled representation learning framework that transforms and filters high-dimensional voice data to identify and contain sensitive attributes at the edge prior to offloading to the cloud. We evaluate EDGY's on-device performance and explore optimization techniques, including model quantization and knowledge distillation, to enable private, accurate and efficient representation learning on resource-constrained devices. Our results show that EDGY runs in tens of milliseconds with 0.2% relative improvement in "zero-shot" ABX score or minimal performance penalties of approximately 5.95% word error rate (WER) in learning linguistic representations from raw voice signals, using a CPU and a single-core ARM processor without specialized hardware.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
