Model-Based Testing for Audio Devices
How We Test on Hardware Early
Firmware can change quickly during development. When it does, tests are prone to becoming bottlenecks – this is especially true when you're validating a variety of audio devices. A single routing change in firmware can cost a significant amount of debugging and guesswork. We needed a way to de-couple test scripts from firmware implementation.
This talk shows how our solution makes tests easier to maintain across multiple audio hardware projects. We'll walk through how our model-based test approach encapsulates signal paths and SKU-specific behaviour. We’ll detail how we achieved this by using a graph network model that runs alongside the device under test (DUT), computing expected signal propagation at run time. In our tests, every audio measurement is tested against acceptance criteria informed by the model.
This enables shared and abstracted tests that serve disparate projects, with scripts that read like checklists. We'll show real project examples, and demonstrate how this approach makes signal paths visible.
If you find that your test scripts are harder to maintain than the firmware they test, this talk is for you.
Max Bailey
Software Developer in Test
Focusrite
Max Bailey is a Software Developer in Test at Focusrite, specialising in test automation for professional audio hardware. With multiple years of experience across audio engineering and quality assurance, he has developed a broad understanding of testing strategies for audio devices. He is passionate about finding ways to 'shift-left' audio hardware tests to maximise the impact of automated testing.