Workshops

Texas Instruments

Real-Time DSP: A Hands-On Tutorial Using the OMAP-L138/C6748 LCD

Presented by: Thad B. Welch (Boise State University) and Michael G. Morrow (University of Wisconsin - Madison)

Monday, May 27, 2013 9:00AM - 4:00PM (lunch included) - VCC Room 114/115

Register at www.ti.com/icassp2013registration

Description

This full day interactive workshop will introduce the use of the OMAP-L138/C6748 LCDK hardware and software kit providing a comprehensive introduction to real-time DSP with an emphasis on how to incorporate these topics into an academic course.

The OMAP-L138/C6748 LCDK is a low-cost development system utilizing TI's dual-core, ARM + DSP OMAP-L138 processor. The TI University Program recommends this as one of the DSP development systems that will be a successor to the C6713 Floating Point DSP Starter Kit with increased performance and additional teaching options. Participants of the workshop will perform hands-on lab exercises concentrating on programming the C6748 DSP core, WinDSK and Code Composer Studio.

During the workshop, the opportunity to discuss the implementation of hands-on DSP teaching will be led by the two experienced and dynamic academic authors/ instructors who have taught many educator to educator based IEEE and ASEE workshops on this topic.

Topics will include:

  • Code Composer Studio (CCS) installation verification
  • RT-DSP introduction and reference text discussion
  • OMAP-L138/LCDK out of box experienced
  • winDSK8 installation and verification and examples
  • Creating a project ... talk through
  • FIR filter and Signal Generation projects
  • Creating a project ... sinusoidal signal generation
  • Advanced topics discussion

Eligibility and Registration

Space is limited so please register early. TI will work to accommodate registrants and preference is given to academic instructors.

Attendees for the full workshop will receive the LCDK board, full version of Code Composer Studio Software, as well as copies of the teaching materials.

MathWorks Workshops

Join Dr. Houman Zarrinkoub as he demonstrates the latest features in MATLAB for modeling and simulation of DSP and communications systems. Please register at www.mathworks.com/icassp2013.

Real-Time DSP System Simulations in MATLAB

Monday, May 27, 1:00 - 2:30 PM - VCC Room 111/112

In this session, we show how you can use MATLAB to develop real-time DSP algorithms and test benches with the latest features in DSP System Toolbox. We showcase an acoustic tracking system that uses the acoustic sensors in Microsoft Kinect.

Through demonstrations, you will learn how to:

  • Use a library of efficient algorithm components (System objects) to develop your real-time DSP IP
  • Automatically generate C code to accelerate simulation or to integrate your design with other software tools
  • Easily discover and interface to your signal processing hardware
  • Test your design with real-time streaming data

Accelerating Communications System Simulations in MATLAB and Simulink

Monday, May 27, 3:00 - 4:30 PM - VCC Room 111/112

In this session, you will learn various techniques you can use to accelerate your communications system simulations in MATLAB and Simulink. We showcase a series of six types of optimizations applied to accelerate the simulation of a 4G LTE control channel processing algorithm. We start with a baseline algorithm and through successive profiling and code updates introduce the following optimizations:

  • Better MATLAB serial programming techniques (vectorization, preallocation)
  • Use of System objects
  • MATLAB to C-code generation (MEX)
  • Parallel computing (parfor, spmd)
  • GPU-optimized System objects
  • Rapid Accelerator mode for simulation in Simulink

Through demonstrations we also show that you can further accelerate your simulations by combining two or more of these techniques.

Please register at www.mathworks.com/icassp2013.

About the Presenter:

Houman Zarrinkoub, Ph.D.
Houman is a senior product manager responsible for Communications System Toolbox. He joined MathWorks in 2001 as a development manager for signal processing products. Prior to MathWorks, he spent six years at Nortel Networks as a research engineer specializing in speech processing for wireless systems. He holds a B.S.E.E. from McGill University and an M.S.E.E. and a Ph.D. from the Institut Nationale de la Recherche Scientifique, Universite du Quebec.

Nuance

Registration is closed.

Network based Speech Enhancement at Nuance Communications

Speaker: Mahesh Godavarti

Monday, May 27, 1:00 - 2:30 PM - VCC Room 111/112

Description: Network based speech enhancement (VQA) at Nuance has a long history spanning more than ten years. Network based enhancement is important for wireless providers in cases where the presence of speech enhancement on end-devices is not guaranteed. In this work, we give a high level overview of multiple research breakthroughs that enabled VQA to achieve a leading position in network based voice processing. Highlights include innovations that went into noise reduction, level normalization, voice activity detection, speech intelligibility improvement and various other algorithms that comprise Nuance VQA that made the algorithms cheap but very effective. The work will also cover speech enhancement work undertaken to improve speech quality perception in specific customer networks. What will also be covered is some of the challenges signal processing engineers will have to overcome to develop an effective suite of algorithms. We conclude with a live demonstration of the product based on the latest advancements that allows Nuance to process coded bit-streams without introducing a tandem coding step arising from decoding and re-encoding of speech.

Speech enhancement front-end for far-talk applications in the home environment

Speaker: Markus Buck

Monday, May 27, 3:00 - 4:30 PM - VCC Room 111/112

Description: For speech applications in the home environment room acoustics are a critical issue. Due to large distances between speaker and microphones the direct sound signal component becomes small compared to interfering signal components. For this reason, there are several challenges that have to be addressed simultaneously in a speech enhancement front-end: robust echo cancellation to get rid of the loudspeaker feedback, suppression of late reverberation, and beam forming to suppress interfering sounds coming from other sources.

In this contribution a speech enhancement system is presented and demonstrated that enables speech applications for far-talk conditions. Typical use cases are ASR applications for the TV or the audio entertainment system, where a microphone array is integrated.