Project Proposal Audio Compression Variants Doris Jean ELEN EE6820 djean95@yahoo.com Hi everyone. My name is Doris Jean. I am a CVN student in the EE master’s program. The purpose of my presentation is to give an overview of my project on MPEG-1 Audio Compression Algorithms Variants.

Outline Overview of Data Compression Project Scope Project Goal Closing Thoughts My presentation will start with a review of data compression fundamentals. Then will give an overview of the project’s purpose and scope. I’d like to also go over some IEEE references that I would like to acquire, and conclude the presentation.

Data Compression Fundamentals Needed to accommodate memory constraints in digital communication systems. Two types of compression techniques; Lossless Used for applications that require the exact replica of the original data after decompression. Lossy Used for applications that don’t require decompressed data to be restored exactly as the original. Compression efficiency is measured by the Compression Ratios. Lossy techniques have greater ratios than Lossless The need for efficient digital signal/data compression has become crucial in recent years due to channel capacity and memory limitations in digital communication systems. There are two types of data compression techniques; lossless, which is used for applications that require the exact replica of the original data after decompression. For example, data compression for text files uses the lossless method to avoid loss of characters in the uncompressed text. The lossy data compression technique in turn is used for applications that don’t require that the data be restored exactly as the original. Both of these data compression techniques perform efficiently in their respective type of digital data applications. The compression ratio for Lossless technique ranges from 2:1 to 8:1. The lossy technique typically can achieve higher compression ratios.

Data Compression Fundamentals (Cont’d) Three categories of compression techniques: Entropy coding Uses lossless compression techniques Source Coding Uses lossy technique Include Sub-band coding, Sub-sampling, Bit Positioning, DCT, FFT, DM and DPCM as compression algorithms Hybrid Coding Uses a combination of Entropy and Source coding Used for MPEG-1 audio compression algorithms There are three categories of compression techniques Entropy coding uses a lossless process Source Coding uses a lossy process. Sub-band coding, Sub-sampling, Bit Positioning, DCT, FFT, DM and DPCM are the compression techniques that fall under this category Hybrid Coding uses a combination of Entropy and Source coding. This compression coding category is used for audio/speech and video data compression systems by using the following compression algorithms; JPEG, MPEG, H.263, and other proprietary algorithms.

Project Scope MPEG-1 Audio MPEG-1 has three layers: layers I, II, III Audio compression algorithm that is based on sub-band coding Widely used in digital audio systems MPEG-1 has three layers: layers I, II, III These layers differ by their signal reconstruction accuracy and by the computation complexity involved in the psychoacoustic model. MPEG 1 Layer III, also called MP3, has better audio signal reconstructions than Layer I or II. However its computation complexity is more involved in layer III and it requires more processing time. MPEG 1 (Moving Picture Expert Group) is an audio signal compression standard / algorithm that is based on sub-band coding compression method, and is widely used in digital audio systems. In the sub-band coding method, the digital signal is divided into multiple sub-bands of frequency by multi-rate filters, and then quantized by allocating bits to each sub-band signal with respect to its amplitude. In addition, MPEG 1 Audio has three layers, layers I, II, and III. These MPEG1 layers differ by their signal reconstruction accuracy and by the computation complexity involved in the psychoacoustic model. For example, MPEG 1 Layer III, also called MP3, has better audio signal reconstructions than Layer I or II. However its computation complexity is more involved and it requires more processing time.

Project Scope (Cont’d) MPEG-1 audio compression method has two mean drawbacks: Has lossy characteristics at high compression ratios Computations involved in the psychoacoustic model part of the compression process causes time delay My project will focus on the MPEG-1 computations drawbacks The MPEG-1 audio compression method has two mean drawbacks; First it has a lossy characteristic at high compression ratio, where part of the original audio signal can be lost when restored or decompressed. Second, the computations involved in the psychoacoustic model causes a time delay, which is not favorable in real-time audio transmission.

Project Scope (Cont’d) MPEG-1 Compression Process Frequency Division Quantization Entropy Encoding (Subband Coding) Compressed Audio Signal Input Audio Signal Psychoacoustic Model Controls Audio compression process is depicted on this slide. It depicts the steps involved in compressing audio signals.

Project Goal Investigate the time delay involved in MPEG-1 Audio layer III compression algorithms Compare the delay to that of MPEG-1 layer II Propose design modifications to reduce the delay in Layer III algorithms Value added Achieving time delay reduction in MPEG-1 Audio compression algorithms The purpose of my project is to investigate and compare the time delay involved in MPEG-1 Audio layer II and III algorithms in the psychoacoustic model computations. Some design modifications considerations will be provided on how MPEG 1 Layer III can improve its delays during computations Value added: - Achieving reduction in time delay will maximize MPEG-1 Layer Is performance efficiency in real-time systems like the internet

Closing Thoughts I would like to acquire these two IEEE articles in reference to my continued effort on the project. * MPEG Digital Audio Coding, P. Noll, pp. 59-81. * MPEG Digital Video-Coding Standards, T. Sikora, pp 82-100 I look forward to your comments and feedback. -Thank you