Signal Processing
Mobile Smartphone Communication Technology
- Prototyped a mobile smartphone device-to-device communication platform using Matlab.
- Converted Matlab simulation to working DSP code included implementing custom Fourier Transform.
- Developed a working hardware solution which consisted of USB-drivers and dual real-time 48kHz audio decoder in 10 MIPS.
- Aided in design, layout and bring up of TMS320C5505 processor boards, including custom board tests.
- Supported a fixed-point Software Development Kit, which runs on various operating systems, including Windows, Linux, iOS and Android.
- Debugged Windows, Linus, iOS and Android implementation to help reduce CPU load and improve performance.
Speed Info
- Designed a non-linear phase-adjustable Hilbert transform and FFT analysis package from start to finish on a TMS320VC5510 processor. Design phase included analysis of stereo signal generated to produce worst-case scenarios as well as real-world capture of data from prototype hardware. Initial analysis consisted of over a dozen Matlab scripts with graphical output.
- Conversion of Matlab simulation to working DSP code included implementing a non-linear phase-adjustable Hilbert Transform and complex FFT.
- Final system consisted of a control task, a signal-processing task, and an input/output task running in real-time using DSP BIOS, Reference Framework 3.
- Project was completed and demo-ready on a Spectrum-Digital DSK in two months.
Audience
- Optimized a noise suppression algorithm on a TMS320VC5416 and a TMS320VC5509A. The algorithm was a proprietary noise cancellation algorithm running in floating point in windows. Port to the embedded DSP platforms allowed the customer to demo their algorithm in real-time on several DSKs.
- Created several demo systems using the EVM5509A and a custom hardware interface including LED displays, and switch inputs, as well audio in and audio out.
- Pacificware's work on these projects led to Audience being awarded the 2008 Wall Street Journal Innovation Award.
A Free Codec For Free SpeachFonly Institute
- Ported Speex to a TMS320VC5502 to compress 8kHz voice to 4kbps.
- Created a custom hardware interface to a Micron MT9T001 ½-Inch 3-Megapixel CMOS Digital Image Sensor. Wrote I2C and DMA software to communicate with the sensor in real time.
- Ported several image processing algorithms to C5502, including edge detection, image rotation and image projection.
- Created printer drivers and a voice equalizer circuit to adjust for speaker frequency response.
- Put together a complete demo system on a Spectrum Digital EVM with a camera, printer, speaker and keyboard.
Nuvation
- Developed a board support package for a TMS320VC6446 (DaVinci) DSP with 256Megs memory and a dual camera interface with I2C control.
- Developed a board support package for a TMS320VC6416 coprocessor.
- Optimized several video and audio algorithms on the 64+ core of each processor.
- Developed a video processing algorithm with Video scaling, image rotation and On-Screen-Display for a robotic application using a TI DM642. The application used a TI TCP/IP stack and video display and performed real-time sensory input processing.