Your Password Is Music To My Ears: Cloud Based Authentication Using Sound

This paper details the research In progress into identifying and addressing the threats faced by voice assistants and audio based digital systems. The popularity of these systems continues to grow as does the number of applications and scenarios they are used in. Smart speakers, smart home devices, mobile phones, telephone banking and even vehicle controls all benefit from being able to be controlled to some extend by voice without diverting the attention of the user to a screen or having to use an input device such as a screen or keyboard. Whilst this removes barriers to use for those with accessibility challenges like visual impairment or motor skills issues and opens up a much more convenient user experience, a number of cyber security threats remain unanswered. This paper details a threat modelling exercise and suggests a model to address the key threats whilst retaining the usability associated with voice driven systems, by using an additional sound based authentication factor.