Developed enhanced audio dataset support for the Model Zoo Test Generator within the ROCm/AMDMIGraphX repository, focusing on expanding test coverage and improving reliability for audio model validation. Leveraged Python and shell scripting to automate test execution, update dependencies such as ffmpeg and torchcodec, and streamline CI/CD workflows. Introduced a summary log for test results, which improved traceability and facilitated more efficient QA reporting. Enhanced the test runner to provide clearer failure outputs, enabling faster root-cause analysis and reducing debug time. The work emphasized robust testing practices and dependency management, contributing to more stable and maintainable model validation processes.