Abstract

Objective

To determine the amount of data compression that is acceptable in different genres of digital audio music recordings.

Introduction

Listening to music is one of life's great joys, but for much of human history, music has only been available to people in a live format. All of that changed in the late 1800's with the development of the first audio recording device, which saved and played back the song "Mary Had a Little Lamb." So it wasn't exactly Justin Timberlake, but it was thrilling to hear nonetheless.

How audio data is saved has changed dramatically over the last century, and even in the last two decades. You probably can't imagine living without an MP3 player, or surfing the Internet without audio, but those are very recent inventions. MP3 is a digital audio compression algorithm that was developed to make digital audio files smaller and easier to move around. The MP3 algorithm takes an uncompressed digital audio recording, often in a standard format known as WAV, and shrinks it down to a smaller size. It works by selectively taking out those frequencies from the original audio recording that the human ear can't hear very well. The resulting MP3 recording is typically ten times smaller than the original and, for most people, sounds close enough to the original to provide a good musical experience.

As an example, a standard, digital, "CD-quality" audio recording is sampled 44,100 times each second, and then stored in 16 bits of data. Multiplying 44,100 by 16-bit samples yields 705,600 bits per second of data for one channel. Stereo requires two channels, so for each second on a CD recording, you are storing two times 705,600 bits per second or 1.4 million bits per second! Recording and saving just one minute of a song requires approximately 10 megabytes (MB) of data space! So if you wanted to save just one of your standard, 70-minute music CD's (70 minutes was chosen as an industry standard), you would need 700 MB of data space!

The MP3 data-compression algorithm fixes that by compressing the recording and eliminating those frequencies that most humans have trouble hearing. The frequencies that are removed do not have to be stored, and therefore, the size of the MP3 file is smaller than the original. How much smaller the MP3 file is from the original recording depends upon you. You choose how much of the original recording you want to keep or remove by adjusting the bit rate, the number of bits per second with which you encode your MP3 file.

The bit rates that are typically allowed for encoding with most software range from 96 to 320 kilobytes per second. The higher the bit rate, the closer you are in fidelity to the original recording. If you want a compressed recording that sounds about like what you hear on the radio, you can choose a bit rate of 128 kilobytes per second, which is considered medium-range quality. In this music science fair project, you'll put your listening and encoding skills to work and see, for different kinds of music, what bit rates are acceptable (are close enough in fidelity to the original to provide a good musical experience), and what bit rates make music sound like singing chipmunks. How low can you go?

Terms, Concepts, and Questions to Start Background Research

• Audio
• MP3 player
• Digital
• Compression
• Algorithm
• WAV
• Frequency
• Sample
• Bit
• Compression
• Bit rate
• Encode
• Fidelity
• Pitch
• Tempo
• Genre

Questions

• How does the digital audio compression algorithm, known as MP3, work?
• Why is MP3 useful?

Bibliography

This source describes what MP3 does and why it is useful:

• Brain, M. (2008). How MP3 Files Work. Retrieved October 24, 2008, from http://computer.howstuffworks.com/mp3.htm

This source describes the history of the MP3 algorithm:

• Wikipedia Contributors. (2008, October 23). MP3. Wikipedia: The Free Encyclopedia. Retrieved October 24, 2008, from http://en.wikipedia.org/w/index.php?title=MP3&oldid=247148358

This link provides free, open-source software with the capability for creating MP3 files with different compression bit rates:

• Audacity Developer Team. (2000, May). The Free, Cross-Platform Sound Editor. Retrieved October 16, 2008, from http://audacity.sourceforge.net/

This source describes how to save MP3 files using free open-source software:

• Boutell.com. (2008). How Do I Record an MP3 File? Retrieved October 24, 2008, from http://www.boutell.com/newfaq/creating/recordmp3.html

Materials and Equipment

• Personal computer with Internet access
• Audacity software, a link to the free download of this software is provided in the Bibliography
• Music CD's or .WAV files (5 songs)
• Lab notebook

Experimental Procedure

1. Select five songs to test from your CD's or .WAV files. Try to choose songs with different pitch and tempo qualities. For example, select one song that has a lot of low, deep frequencies, another that has high-pitched frequencies, one with a slow beat, one with a fast beat, and one with a mix of qualities. Or try song titles from different genres of music, such as classical, hip-hop, jazz, pop, bluegrass, electronic, folk, reggae, blues, or rock.
2. Select three bit rates of encoding that you would like to test (low, medium, and high) and enter the values you have chosen in a data table, like the one below, in your lab notebook.
3. Download and run the Audacity software and capture a sample of one of your chosen songs that is 10-30 seconds long. Record the length of time of your sample, so that you can use the same length of time for your other trials.
4. Encode the sample with the low-compression bit rate you have chosen. Save the MP3 recording you have created.
5. Encode the sample with the medium-compression bit rate you have chosen. Save the MP3 recording you have created.
6. Encode the sample with the high-compression bit rate you have chosen. Save the MP3 recording you have created.
7. Repeat steps 3-6 for your remaining song titles.
8. Listen to the first song from your CD or .WAV file.
9. Listen to the low-compression MP3 version of the song. Is it acceptable to your ear and a good music experience? Write down your observations in your data table.
10. Listen to the first song from your CD or .WAV file again.
11. Listen to the medium-compression MP3 version of the song. Is it acceptable to your ear and a good music experience? Write down your observations in your data table.
12. Listen to the first song one more time from your CD or .WAV file.
13. Now listen to the high-compression MP3 version of the song. Is it acceptable to your ear and a good music experience? Write down your observations in your data table.
14. Repeat steps 8-13 for the remaining songs.
15. Examine your data table. Did the amount of acceptable data compression change with each song? What is different about the songs that were able to take a low-compression bit rate and still be acceptable? What is different about the songs that required a high-compression bit rate to be acceptable?

Data Table: Acceptable and Not Acceptable Musical Experiences

Variations

• Using the Audacity software, compare the frequency spectra of an original song recording with the low-, medium-, and high-compression MP3 versions. How did the bit rate affect the spectra? Compare the spectra with a chart that shows the threshold of hearing for humans to determine the frequencies at which humans will experience the greatest loss in fidelity for this particular song at each bit rate.
• Data can also be stored in 8, rather than 16, bits of data during sampling. This reduces fidelity, but saves space. Using Audacity software, compare the quality of music when it is sampled and stored on 8 bits, with the quality of music when it is sampled and stored on 16 bits.

• For more science project ideas in this area of science, see Music Project Ideas.

Credits

Kristin Strong, Science Buddies

Edited by Peter Boretsky, Lockheed Martin

Last edit date: 2008-12-05 10:27:00