SUMMARY
1. INTRODUCTION
2. OBJECTIVES
3. METHODOLOGY
4. SPECIFICATIONS
5. PROJECT MANAGEMENT
5.1 Resources
5.2 Schedule of Project Milestones
6. TECHNICAL AND PRODUCT REFERENCES
Appendix A -MPEG-2 introduction.
Appendix B -Team members resumes.
Appendix C - List of acronyms used in the report.
Currently the most common media for storing audio is the compact disc. New formats of media called mini-disk and DCC have been released with in the past year which use different types of lossy compression to reduce the bandwidth of digital audio while tr ying to retain the qualities of digital audio. However, with the latest expansion of multimedia and networking onto the computer, there is a growing need for integrating of the previously mentioned media with the computer. In the future it is easy to en vision on-line music databases and stores where you can sample and purchase music. The computer will be the center for controlling all communications and media within the home. Our design project is to contribute to this trend of integrating multimedia w ith the computer.
The main problem in this integration is that CD-quality digital audio requires a great amount of bandwidth and storage-spaces. For example, each compact disk can hold approximately 700 megabytes of digital information which corresponds to 70 minutes of m usic. With the current technology it is possible to sent and store music on computer, but it is not viable because the cost of the disk-space and high-bandwidth communication lines.
One possible way to overcome this problem is by compressing the audio Bitstream. Compression of the audio Bitstream will reduce the bandwidth and size of the stored data. There are many software compression utilities but they require much of computer res ources just for conversion. Due to the development of powerful signal processors it is possible to obtain a real-time audio compression and decompression. Our design project is using this approach by implementing a `MPEG-2 layer 3' decompression module in a regular PC sound-card.
The most used form of digital audio is compact disc digital audio. The analog audio is sampled at 44.1 kHz, where the samples are represented with 16 bit precision. A 5 minute song uses approximately 50 MB (megabytes) of data. This amount of data is expen sive to store, and impractical to send across current networks. This project presents a design of a specific PC sound card that uses on-board compression utility and reduces the need for large amounts of storage space.
The development of DSP chips have made it possible to implement algorithms which will enable a high compression rate. The algorithms currently used are based by psychoacoustic models used to remove information describing sounds which the human ear cannot recognize. The most effective algorithms can reduce the amount of data by 92% with little or no audible loss of sound quality. Since it is a relative new research area and there are a number of different algorithms, after researching the availability an d trends of a few compression standards, we chose to implement a MPEG-2 layer 3 codec which gives the best results for music compression. Depending on the encoder this the rate of compression is on average 1:12. A short introduction to this standard is pr esented in Appendix A.
The implementation of the sound-card will use a commercial MPEG-2 layer 3 decoder. There are few commercially available decoders on the market today and the number is increasing rapidly. The reason we are only implementing the decoder is the complexity an d price of current real time encoders. However, the project will include some kind of compression utility using software package, so that the card will enable recording and storing music in compressed MPEG format and playing this files with real-time deco mpression.
The major engineering objective for our project is to construct a sound-card which will be connected to an ISA bus. The sound-card will use a MPEG-2 layer 3 decoder chip to transform the MPEG Bitstream from the ISA bus to a 16 bit 44.1 kHz Bitstream. The converted Bitstream will then be processed by a DAC (digital to analog converter) to put the sound back into our analog world. There will also be an A/D converter for recording audio signals to disk. These are the primary objectives that will be accompli shed:
- Implement D/A and A/D converters for a 16 bit 44.1 kHz Bitstream.
- Obtain a MPEG decoder.
- Augment MPEG decoder with the D/A module and the ISA bus.
- Write device driver to interface the module with the computer.
The approach taken is to decode MPEG audio in real-time without relying on the CPU and a software decompression algorithm. This will be a major benefit because the CPU's load will not be affected by the complex decompression algorithm and will be able to process other commands. A flow chart on how this system will operate is in figure 1.
In the construction of our sound card we will begin by implementing the DAC (digital to analog converter) and ADC. Using the Linux operating system on a PC with an ISA bus we will interface the converters. We chose this environment because we will have t o write a device driver for the sound-card to interface with the computer, and all the source code for the kernel and other device drivers are available on the net. Ones the device driver is written, a simple command like: cat soundfile > /dev/sound-c ard will allow the raw digital audio to be transferred into the analog domain. By this time the module should work as a regular sound card without any compression options.
Our next step will be to implement the MPEG decoder into the chip. The decoder will be able to recognize the MPEG headers and other information needed to convert the MPEG bitstream to the 16 bit 44.1 kHz bitstream going to the DAC. Also, the device drive r should accomodate the MPEG chip. Simple block diagram of the design is shown on figure 2.
The proposed MPEG audio card will have the following operating characteristics:
- Interface with an ISA bus (A device driver written in c for the linux operating system will control the communication between the computer and sound card.)
- Decode a MPEG-2 layer 3 Bitstream to a 16 bit 44.1 kHz Bitstream.
- Convert a 16 bit 44.1 kHz Bitstream to analog.
- Convert an analog input to a 16 bit 44.1 kHz Bitstream.
5.1 RESOURCES
Our team consists of two members: E.E., Kire Filipovski, and a C.P.E., Mike Schwankl. We are sharing most of the work, but there is some tasks that will be assigned separately because of special experience as stated in our resumes in appendix B. A pie c hart depicting responsibilities is in figure 3. Most of the software and the interfacing (device driver) work will be done by Mike Schwankl and most of the hardware (breadboarding/circuit testing) will be done by Kire Filipovski.
5.2 SCHEDULE OF PROJECT MILESTONES
Figure 4. ing the time-line of our design project. It lists the main packages of work and production versus time.
MPEG-2 introduction
MPEG is ``Moving Picture Experts Group'', working under the joint direction of the ISO (International Standard Organization) and the IEC (International Electro-technical Commission. MPEG-2 is constructed of 3 parts.
1-Video describes compression of video signal
2-Audio describes compression of audio signals
3-System describes synchronization and multiplexing of video and audio.
MPEG audio coding describes the compression of audio signals using high performance perceptual coding schemes. It specifies a family of three audio coding schemes called layer-1,2,3. The major difference between these three layers are (from 1 to 3)
complexity increases (mainly true for the encoder)
overall codec delay increases
performance increases (sound-quality per bit rate)
The standard supports sampling rates of (16, 22, 24, 32, 44.1, 48) kHz and bit rates ranging from 32 kbps-320 kbps (Kilobits/Second) for layer 3.
List of acronyms use in the report:
CD Compact Disk
DSP Digital Signal Processor
GB Gigabytes
ISO International Standards Organization
kHz Kilo Hertz
MB Megabytes
MPEG Moving Picture Expert Group