Case Study: Squeezing Big Algorithms into Small Power Budgets Enables Mobile Audio Quality

Submitted by BDTI on Tue, 09/29/2015 - 22:00

Today's smartphones are technological marvels that deliver an extraordinary range of capabilities from GPS-based navigation to sophisticated photography. But sometimes we just want to make a phone call. And particularly when we're on the move, who hasn't struggled to hear the other party or to be heard on a mobile call?

Realizing the importance of intelligible phone calls – not to mention a strong need to differentiate their products – smartphone manufacturers are incorporating increasingly sophisticated speech and audio processing algorithms into their devices. Of course, sophisticated algorithms can consume a lot of processing power. And while there's plenty of processing power available in modern smartphone CPU, the goal is to use as little of it as possible to maximize battery life. This is especially important in use cases like phone calls, for which consumers reasonably expect many hours of use on a battery charge. In general, there are two keys to enabling such algorithms to fit into mobile phones: implement them on the energy efficient DSP core present in most mobile phone SoCs, and optimize the code to take advantage of the DSP's instruction set and pipeline.

Recently an audio algorithm company developed a sophisticated new ambient noise reduction algorithm. Smartphone makers loved how the algorithm sounded, but weren't thrilled with the requirement of 400 MIPS of processing performance. The algorithm supplier engaged BDTI to quickly shrink its algorithm's processing resource footprint. BDTI's engineers profiled the algorithm to identify hot spots, then optimized key functions and operations to reduce computation and data memory traffic while maintaining high fidelity. In short order, the computation requirements were reduced to roughly 100 MIPS, yielding minimal power consumption on the target DSP core.

Do you have a big algorithm that you need to squeeze into a small cost or power consumption budget? If so, BDTI can help. BDTI creates highly optimized implementations of digital signal processing algorithms for CPUs, DSPs, GPUs, FPGAs and other specialized processing engines.

Contact Jeremy Giddings at +1 925 954 1411 or for a confidential consultation to find out how BDTI can quickly transform your algorithms to meet tough processor performance and power consumption goals.

Add new comment

Log in or register to post comments