Article Addendum: 'Hearing aid: Auditioning high-resolution surround-sound compression', July 10, 2003

This'll be a 'living' document; I'll periodically update it with additional evaluation results on both existing and new codecs, plus additional links that I come across. So occasionally check back, and drop me an email with any feedback. Thanks!

July 10, 2003

Screen Shots

It's unfortunately going to take me far longer than I originally anticipated to capture and JPEG-transform the vast breadth of screen shots that I promised in the 'Surf the Web-site WAVs' sidebar to my article. In the interim, I thought it would at minimum be helpful if I provided you with full-size versions of the article's Figures 2 through 6, so you could more clearly discern the differences between them. Below, you'll find links to the graphics, in both JPEG and zipped BMP formats.

Figure 2

Figure 2a (JPG)

Figure 2b (JPG)

Figure 2c (JPG)

Figure 2d (JPG)

Figure 2e (JPG)
All Figure 2 graphics (ZIP set of BMPs)
Figure 3

Figure 3a (JPG)

Figure 3b (JPG)

Figure 3c (JPG)

Figure 3d (JPG)

Figure 3e (JPG)

Figure 3f (JPG)

Figure 3g (JPG)

Figure 3h (JPG)

Figure 3i (JPG)

Figure 3j (JPG)
All Figure 3 graphics (ZIP set of BMPs)
Figure 4

Figure 4a (JPG)

Figure 4b (JPG)

Figure 4c (JPG)

Figure 4d (JPG)

Figure 4e (JPG)

Figure 4f (JPG)

Figure 4g (JPG)

Figure 4h (JPG)

Figure 4i (JPG)

Figure 4j (JPG)
All Figure 4 graphics (ZIP set of BMPs)
Figure 5

Figure 5a (JPG)

Figure 5b (JPG)

Figure 5c (JPG)

Figure 5d (JPG)

Figure 5e (JPG)

Figure 5f (JPG)

Figure 5g (JPG)
All Figure 5 graphics (ZIP set of BMPs)
Figure 6

Figure 6a (JPG)

Figure 6b (JPG)

Figure 6c (JPG)

Figure 6d (JPG)

Figure 6e (JPG)

Figure 6f (JPG)
All Figure 6 graphics (ZIP set of BMPs)

Source Clips

Below you'll find links to download ZIP'd sets of the synthetic test clips I created for this project, along with a description of each clip. Due to server storage and bandwidth limitations, I'm providing only the 96 kHz-sampled variants; to downsample them to 48 kHz, you can employ a full-featured program like Cool Edit or Sound Forge, or a more focused conversion utility such as the freeware program SoX.

Combo (ZIP set of WAVs) Frequency masking test

The following half-band frequencies are at approx. -1 dBFS combined peak levels (all channels): 50, 150, 250, 350, 455, 570, 700, 845, 1000, 1175, 1375, 1600, 1860, 2160, 2510, 2925, 3425, 4050, 4850, 5850, 7050, 8600, 10750, 13750, 18775 Hz

The following quarter-band frequencies are at approx. -21 dBFS combined peak levels (all channels): 25, 125, 225, 325, 427.5, 540, 665, 807.5, 960, 1127.5, 1322.5, 1540, 1790, 2080, 2415, 2812.5, 3287.5. 3875, 4625, 5575, 6725, 8150, 10125, 12875, 17137.5 Hz

The following three-quarter-band frequencies are at approx. -21 dbFS combined peak levels (all channels): 75, 175, 275, 375, 482.5, 600, 735, 882.5, 1040, 1222.5, 1427.5, 1660, 1930, 2240, 2605, 3037.5, 3562.5, 4225, 5075, 6125, 7375, 9050, 11375, 14625, 20412.5 Hz

96 kHz sampling rate, 24-bit per-channel sample size, six channels

Right front, left surround and LFE channels are 180 degrees out of phase with left front, right surround and center channels

All frequencies are present over the entire time interval
Combo-20 (ZIP set of WAVs) Frequency masking test

The following half-band frequencies are at approx. -1 dBFS combined peak levels (left front, right surround and center channels) and approx. -21 dBFS combined peak levels (right front, left surround and LFE channels): 50, 150, 250, 350, 455, 570, 700, 845, 1000, 1175, 1375, 1600, 1860, 2160, 2510, 2925, 3425, 4050, 4850, 5850, 7050, 8600, 10750, 13750, 18775 Hz

The following quarter-band frequencies are at approx. -21 dBFS combined peak levels (left front, right surround and center channels) and approx. -41 dBFS combined peak levels (right front, left surround and LFE channels): 25, 125, 225, 325, 427.5, 540, 665, 807.5, 960, 1127.5, 1322.5, 1540, 1790, 2080, 2415, 2812.5, 3287.5. 3875, 4625, 5575, 6725, 8150, 10125, 12875, 17137.5 Hz

The following three-quarter-band frequencies are at approx. -21 dBFS combined peak levels (left front, right surround and center channels) and approx. -41 dBFS combined peak levels (right front, left surround and LFE channels): 75, 175, 275, 375, 482.5, 600, 735, 882.5, 1040, 1222.5, 1427.5, 1660, 1930, 2240, 2605, 3037.5, 3562.5, 4225, 5075, 6125, 7375, 9050, 11375, 14625, 20412.5 Hz

96 kHz sampling rate, 24-bit per-channel sample size, six channels (left front, right surround and center channel WAVs are in the Combo ZIP set)

Right front, left surround and LFE channels are 180 degrees out of phase with left front, right surround and center channels

All frequencies are present over the entire time interval
Half (ZIP set of WAVs) The following half-band frequencies are at near-0 dBFS combined peak levels (all channels): 50, 150, 250, 350, 455, 570, 700, 845, 1000, 1175, 1375, 1600, 1860, 2160, 2510, 2925, 3425, 4050, 4850, 5850, 7050, 8600, 10750, 13750, 18775 Hz

96 kHz sampling rate, 24-bit per-channel sample size, six channels

Right front, left surround and LFE channels are 180 degrees out of phase with left front, right surround and center channels

All frequencies are present over the entire time interval
Half-20 (ZIP set of WAVs) The following half-band frequencies are at near-0 dBFS combined peak levels (left front, right surround and center channels) and approx. -20 dBFS combined peak levels (right front, left surround and LFE channels): 50, 150, 250, 350, 455, 570, 700, 845, 1000, 1175, 1375, 1600, 1860, 2160, 2510, 2925, 3425, 4050, 4850, 5850, 7050, 8600, 10750, 13750, 18775 Hz

96 kHz sampling rate, 24-bit per-channel sample size, six channels (left front, right surround and center channel WAVs are in the Half ZIP set)

Right front, left surround and LFE channels are 180 degrees out of phase with left front, right surround and center channels

All frequencies are present over the entire time interval
Pink (ZIP set of WAVs) Pink noise

All channels are at approx. -2 dBFS peak levels

96 kHz sampling rate, 24-bit per-channel sample size, six channels

Each channel contains a unique noise pattern
Pink-20 (ZIP set of WAVs) Pink noise

Left front, right surround and center channels are at approx. -2 dBFS peak levels

Right front, left surround and LFE channels are at approx -22 dBFS peak levels

96 kHz sampling rate, 24-bit per-channel sample size, six channels (left front, right surround and center channel WAVs are in the Pink ZIP set)

Each channel contains a unique noise pattern
White (ZIP set of WAVs) White noise

All channels are at near-0 dBFS peak levels

96 kHz sampling rate, 24-bit per-channel sample size, six channels

Each channel contains a unique noise pattern
White-20 (ZIP set of WAVs) White noise

Left front, right surround and center channels are at near-0 dBFS peak levels

Right front, left surround and LFE channels are at approx -20 dBFS peak levels

96 kHz sampling rate, 24-bit per-channel sample size, six channels (left front, right surround and center channel WAVs are in the White ZIP set)

Each channel contains a unique noise pattern

I've also obtained permission from Microsoft to redistribute the set of 24-bit, 48-kHz-sampled surround music files that Microsoft supplied me at their August, 2002 Windows Media Reviewer's Workshop, and that I mention in the 'WAV woes' sidebar to my article.

ambient_1 (ZIP) ambient_2 (ZIP) ambient_3 (ZIP) ambient_4 (ZIP)
Latin (ZIP) NewOrleans (ZIP) Rock (ZIP) World_Beat (ZIP)

To download the source material for the Castanets, Guitar-Sax, Keys, Snaps, Tambourine and Triangle test clips used in this project, please visit Arny Krueger's PCABX website where you'll find two-channel WAVs in a variety of sample rates and sample sizes. As discussed in my article, you can transform these clips into variants similar to the ones I employed by first extending them to 30+ second playback duration through copy-and-paste waveform repetitions within Cool Edit Pro or your favorite audio editing program. Then, duplicate the clips’ left front and right front channels to create, respectively, the left  and right surround channels, and blend the left front and right front channels to create the center and LFE channels..

Batch Files

As mentioned in the article, it's possible to run the freeware Windows Media Encoder in script mode. WMAMSCLIPS.BAT is the script file I used to encode the Microsoft-supplied surround music files, after first transforming each of them from six individual WAV files to a single six-channel AVI using Microsoft's Mono to Multichannel WAV Combiner utility (which, being a command line-drive utility, you an also run from a batch file). WMAPCABX.BAT similarly encoded the PCABX files, while WMASYNTHETIC.BAT encoded the synthetic audio test clips that I created. You can download these script files and modify them as necessary to fit your setup; note that as currently coded they must be in the same folder as Windows Media Encoder. Since the Microsoft command line WMA-to-WAV decoder isn't publicly available, I haven't provided the batch files I used when running it. For similar reasons, I haven't supplied the batch files I used when running the DTS encoder and decoder.

August 19, 2003

At the conclusion of this project, I ended up with almost 23 GBytes' worth of AVI, BAT, BMP, EXE, WAV and other files clogging my hard drive, organized in an extensive directory (aka 'folder', for those of you who don't remember DOS!) 'tree'. How to archive them, get them off my hard drive and thereby free up space again? For maximum portability to multiple computers, I wanted to employ optical storage media instead of DSS tape...the sub-GByte storage capacity of CD-Rs made the writeable DVD choice an obvious one. Copying files 4.7 GBytes at a time would have been a tedious process that also would have made it a challenge to preserve the project's original directory structure.

Instead, I decided to use backup software to accomplish the task. The Backup utility included within Windows XP unfortunately doesn't support writing to optical media, only to floppy discs, to archive images on the hard drive, and to tape drives. Backup software such as Acronis's True Image works well, but only on entire drives and partitions, not on selected directories and files within them. I ended up using Stomp's BackUp MyPC, which worked great. Unlike some competitors' products whose reviews I'd researched, BackUp MyPC wrote to my Pioneer DVR-104 (also known as the DVR-A04) at its fastest-possible speed, and I was able to post-backup verify that everything was correctly burned to the multiple DVD-Rs prior to deleting the files off my hard drive. The only disappointing aspect to the software that I discovered is that it didn't seem to compress the WAV files even remotely to the degree that the multimedia-tuned compression algorithms in programs like FLAC, Monkey's Audio, RAR and Shorten can.