Customer: CODEC manufacturer partnering with a major OEM and ODM
Platform: Intel and AMD based laptop models running Windows 11
Summary: Full-featured driver alternative to RealTek with advanced features such as render processing offload on DSP, Hardware Keyword Spotting and Wake-On-Voice. Supports both Alexa and Cortana on Windows. Managed a multi-vendor software stack in tight partnership with other vendors. Included branded installer supporting multiple packages from Intel, AMD, Waves and others. Employed our own automated Continuous Integration and testing framework that ensured that the WIndows HLK tests were passing before each code update was merged.
Customer: Spatial Audio processing algorithm provider partnering with a major OEM
Platform: Laptop running Windows 11
Summary: For this project, we provided a licensed framework that allows algorithm developers to create logoed audio processing (APO) modules for Windows, with driver packages for both HD Audio and USB devices. Algorithm developers can use the framework to insert their own processing code (“your code goes here”) without the need to share proprietary source code with Optimal Sound. A test shell is provided for rapid validation of algorithms prior to the creation of a fully implemented APO.
Benefits to Customer: Rapid prototyping on an existing laptop to win business with an OEM; logoed driver package with installer for both Windows 11 and Windows 10.
Employer: Magic Leap
Platform: ML1 and ML2 Virtual Reality Headsets; Android/Linux; NVIDIA chipset with audio coprocessor; AMD chipset with Cadence DSPs
Application: Full VR Spatial Audio Stack
Summary: Prior to rebooting Optimal Sound, David Roach led the acoustics, audio hardware and audio software teams at Magic Leap. (Four members of that team are now with Optimal Sound.) David and the team architected, designed and implemented a full spatial audio stack, including audio APIs backing algorithms provided by the Magic Leap audio algorithms team. The project included custom hardware and audio I/O processing firmware integration with multiple DSPs. This layer totally replaced Android’s audio flinger, and was implemented to “bare-metal”. The stack was then ported to the more powerful AMD Zen2 CPU and its’ HiFi3 DSP core, in addition to HiFi4 cores dedicated for speaker and microphone pre and post-processing.