ECE 4760: Final Project


Real Time Audio Compression

Cheng Xing (

Rene Zhang (

"Real time CD quality audio compression and transmission via adaptive differential pulse code modulation."

With the emergence of Internet of Things, it has become easier than ever to transmit data across embedded systems. More and more modules are designed to enable fast and reliable wireless transmission of data, such as Bluetooth, WiFi, and WiFi-Direct. A valuable use case of these systems is audio transmission. With the Microchip PIC32 that the ECE 4760 course is using this year, we have enough processing power to transmit high-resolution audio data. However, wireless transmission bandwidth and power consumption are still two major limiting factors. To tackle this issue, we can exploit the fact that it is possible to compress audio data such that the human ear can barely discern the difference between original and compressed version. In real time, we can apply signal processing techniques to reduce the size of audio data without losing a significant amount of audio quality, thus allowing audio to be transmitted under relatively strict transmission bandwidth and power consumption constraints.

As our final project, we have explored a particular compression scheme, Adaptive Differential Pulse Code Modulation (ADPCM). ADPCM is capable of a 4:1 compression ratio, which significantly reduces the transmission size. We have designed and constructed separate transmitter and receiver modules. The transmitter samples audio, compresses the resulting digital data, and transmits it to the receiver over SPI. The receiver then decompresses the data and interpolates it to reconstruct the audio signal. In order to preserve the quality of the original audio signal and isolate the effects of ADPCM, we sample and interpolate with CD quality, i.e. sampling frequency of 44.1kHz and 16-bit resolution. We have also implemented our system with stereo audio in order to simulate realistic data throughput.

Figure 1: BoomBox hardware