Hearing connects us to people. Hearing impairment has a direct impact on quality of life. However, correcting for hearing is a much more complex problem when compared to our other senses, like sight, because our brain interprets sound through a complex set of mechanisms, each with its unique functionality. Hearing aids try to compensate for many of these mechanisms and have difficulty replicating their functionality. Many researchers are actively working on designing the next generation of algorithms to address the problem of hearing loss through hearing aids. These studies are limited in their impact because of custom or commercial platforms used by different groups. Without an accessible open platform and necessary experimental infrastructure and tools, researchers are hindered in their ability to understand the algorithm's efficacy in the field and prevent adoption into widespread use. In order to address this challenge, this research presents how the Open Speech Platform~(OSP), a hardware and software research tool, was designed and developed to address this gap in the hearing aid research community.

First, this dissertation shows that mobile computing platforms are suitable hardware devices for developing and testing hearing aid algorithms in the field. Here we build a proof of concept device on the Qualcomm 410c platform that we both objectively and subjectively show that it meets current best practices set by the industry.

Second, this dissertation outlines important design choices and their impact on the real-time master hearing aid~(RT-MHA) framework, the software at the core of the OSP. A significant challenge in designing the framework for mobile computing platforms is dealing with a mixed real-time system on a multi-core system while sacrificing the least regarding programmability and power. This dissertation starts the design process at the system's core, a.k.a. the operating environment, and then describes how best to balance the workload across the resources available in a multi-core mobile computing platform.

Finally, this dissertation describes and evaluates the two mechanisms we designed to address the two significant challenges in developing hearing aid algorithms using the OSP. The first is the scalability of the OSP. Initially, each algorithm needs to be statically compiled, making changing algorithms difficult. Therefore, we developed a hot-swapping mechanism plus a Rapid API system to address this problem. The second is the lack of debugging after deployment. We developed Live OScope, a debugger for the OSP modeled after the oscilloscope. This dissertation shows that the Live OScope is a multipurpose tool for research.