Audio Comparison With Audacity and Quicktime Compression Codecs
When I originally started this assignment, I began with a recording of a MIDI file
that used SoundFonts. For the sake of interest, my
original comparison has a set of degraded MIDI audio. For this assignment, I decided to use
this benchmark instead, as the audio is of
real instruments. The 30 second audio clip was taken from Evening Falls (Roomfall of Emptiness, Matti
Paalanen) at Jamendo; Electric
Guitar, Jazz Piano, Bass and Percussion. It is a royalty free site.
(benchmark file, 16-Bit Mono @ 44100 Hz)
Audio Compression Codecs
| Filename | Codec | Sample/Bit Rate | Data Rate (kbps) | File Size (kb) |
|---|---|---|---|---|
| Benchmark | -- | 44,100kHz, 16-bit | 705.42 | 2580.48 |
| MP3-128 | MP3 | 44,100kHz, 16-bit | 469.79 | 127.96 |
| MP3-64 | MP3 | 44,100kHz, 16-bit | 234.89 | 63.98 |
| MP3-32 | MP3 | 44,100kHz, 16-bit | 31.99 | 117.34 |
| MP4-150 | AAC | 44,100kHz, 16-bit | 84.21 | 307.75 |
| MP4-75 | AAC | 44,100kHz, 16-bit | 43.38 | 158.54 |
| MP4-30 | AAC | 44,100kHz, 16-bit | 22.06 | 80.63 |
| WMA-128 | Windows Media Audio | 44,100kHz, 16-bit | 506 | |
| WMA-64 | Windows Media Audio | 44,100kHz, 16-bit | 253 | |
| WMA-32 | Windows Media Audio | 44,100kHz, 16-bit | 138 | |
| u-Law | u-Law | 44,100kHz, 16-bit | 352.71 | 1290.24 |
| A-Law | A-Law | 44,100kHz, 16-bit | 352.71 | 1290.24 |
| IMA | IMA 4.1 | 44,100kHz, 16-bit | 176.11 | 646 |
Comments: I find that the MP4 file at 22.06 kbps is a good choice for use in a website as it is of good quality and has a low file size. The MP3 file at 31.99 kbps is also a good choice. Overall, I found the samples to be of the same quality as the benchmark. For each sample, audio compression has provded a smaller file size with no significant loss. This surpised me, as I was expecing more loss.
| Filename | Codec | Sample/Bit Rate | Data Rate (kbps) | File Size (kb) |
|---|---|---|---|---|
| QDesign 32 | QDesign Music 2 | 44,100kHz, 16-bit | 31.76 | 116.34 |
| QDesign 24 | QDesign Music 2 | 44,100kHz, 16-bit | 23.86 | 87.86 |
| QDesign 20 | QDesign Music 2 | 44,100kHz, 16-bit | 19,91 | 72.95 |
| QDesign 16 | QDesign Music 2 | 44,100kHz, 16-bit | 15.79 | 57.85 |
| QDesign 12 | QDesign Music 2 | 44,100kHz, 16-bit | 11.85 | 43.39 |
| QDesign 10 | QDesign Music 2 | 44,100kHz, 16-bit | 9.96 | 36.47 |
| QDesign 8 | QDesign Music 2 | 44,100kHz, 16-bit | 7.90 | 28.92 |
| MACE 3.1 | MACE 3.1 | 44,100kHz, 16-bit | 117.57 | 430.66 |
| MACE 6.1 | MACE 6.1 | 44,100kHz, 16-bit | 58.78 | 215.33 |
| IMA 4.1 | IMA 4.1 | 44,100kHz, 16-bit | 187.37 | 686.34 |
Comments: I find that the QDesign Music 2 Codec provides good audio compressions at 32, 24 and 20 kbps. The quality is good and the file sizes are low (116.34 KB, 87.86 KB and 72.95 KB respectively). I cannot hear a significant difference between the benchmark file and the audio samples derived from it. I believe that the results one gets largely depends on the audio source. If one compares this set of files with the original files I created (MIDI source), there IS a difference in quality. Not all audio takes kindly to compression, and one may not even know compression has been used. We tend to notice it more when it is far too high.
In Assignment 2, we were asked to create a chart showing various audio compression codecs. I have split mine into Audacity and
Quicktime files. I made my benchmark file from a stereo recording I made of the audio at Jamendo, and converted the file to mono.
After I normalized this audio file, I saved all the required samples -- and upon listening to them, didn't notice any significant differ-
ence. For websites, I would choose one of these 2 compressions: MP3 or MP4, and would favor the use of the MP4 files
due to their low file sizes. I would limit the audio file size to under 200 KB. This would mean that the download time would be less than
30 seconds for a 56 KB modem and even less for high speed internet connections.
Unlike my first go at this assignment, this time I have saved the differnet codecs using different qualities (e.g. 128 kbps, 64 kbsp, 32 kbps, etc). This variance has allowed for lower file sizes. For websites, use of audio with low file sizes can make all the differ- ence with respect to download time. Along with all the other things that a browser must download, the audio is included, especially those files that are embedded. Every resource on a website takes a finite amount of time to download -- and the lower the file size, the better. The quality of the audio for my sample files was good, but there is still some loss -- though it is not very notice- able the first time the audio is listened to. I'm sure if I listened more closely, I'd notice some differences. My hearing is not as good as it used to be. The sampled audio from the benchmark is quite clear, but I know there ARE differences.
Unlike my first go at this assignment, this time I have saved the differnet codecs using different qualities (e.g. 128 kbps, 64 kbsp, 32 kbps, etc). This variance has allowed for lower file sizes. For websites, use of audio with low file sizes can make all the differ- ence with respect to download time. Along with all the other things that a browser must download, the audio is included, especially those files that are embedded. Every resource on a website takes a finite amount of time to download -- and the lower the file size, the better. The quality of the audio for my sample files was good, but there is still some loss -- though it is not very notice- able the first time the audio is listened to. I'm sure if I listened more closely, I'd notice some differences. My hearing is not as good as it used to be. The sampled audio from the benchmark is quite clear, but I know there ARE differences.