Hardware And Software Reviews
Ultramaximize Your Digital Audio:
Read G. Burgan
I was having a coffee break in the snack bar of the Michigan Technological University Memorial Union building when John strode over to our table. "I can't listen to WGGL in my truck!" The words were accusing, suggesting we had a glaring problem.
John was a professor in the electrical engineering department and a loyal public broadcasting supporter. At the time I was the director of Tech's public radio station. As the only 100,000 watt station in the area, I knew it couldn't be our signal strength that was creating the problem. Since Houghton, Michigan, lay at the bottom of a valley , we had our share of multi-path, but no more so than any other broadcaster in the area. So what was our problem?
A broad smile broke out on John's face as he watched me ponder his statement. "Your dynamic range is so great that the noise in my truck makes it impossible to listen without turning the sound so high that the crescendos deafen me," he blurted out, the smile broadening to a full laugh.
John was right. While all of the commercial stations in our area were pumping up their modulation with various limiting devices, we prided ourselves on maintaining as much of the original dynamic range as possible. But there were downsides. His experience was one. It also meant that in some portions of our coverage area, stations with a lot less power than ours sounded louder. "You have 100,000 watts? It sure doesn't sound like it?!"
And when in stereo, listeners at the outer edge of our coverage area often had a marginal signal. In the 70's we regularly dropped our stereo subcarrier during broadcasts of mono programs like "All Things Considered" in an effort to boost our signal in those areas.
Purists have argued for decades over the advantages and disadvantages of applying various forms of limiting to a broadcast signal. From the 1940's on, manufacturers have made equipment to boost a station's perceived output by maximizing modulation through various forms of limiting. With the rapid increase in the number of metropolitan stations in the past several decades, there's probably not a single station that doesn't use some form of limiting in an effort to stand out in the crowd.
In some cases the result is hardly noticeable. In others, the altered sound has become an area trademark, a symbol of pride. In still others it's been a blot, creating an ugly sound to all but a tone deaf management. Like it or not, limiting is a part of broadcasting.
Now you can apply limiting to your program material before you broadcast it -- digitally. K.S.Waves Ltd. has produced a software add-on plug-in for Sonic Foundry's Sound Forge 3.0 that allows you to apply sophisticated peak limiting to any sound file in the waves format. It has other features that will improve sound files that are converted to shallower bit modes.
The problem with peak limiting is that it often creates distortion or a signal so compressed that it calls attention to itself. Many analog peak limiters are unable to deal with the wide variety of peaks and either allow some peaks to slip beyond the set level or introduce distortion in an attempt to suppress an unusually strong peak.
The L1-Ultramaximizer allows you to apply peak limiting in a way that rigidly maintains the signal at precisely the level you select without fear that some portions of the signal will slip through or distort. According to its designers, "The L1-Ultramaximizer avoids the possibility of overshot by utilizing a lookahead technique that allows the system to anticipate and reshape signal peaks in a way that produces the bare minimum of audible artifacts."
Does it work? You bet. The measure of a product's success is 1. How well it performs the function for which it was designed, and 2. How easy it is to use. In both cases, the L1-Ultramaximizer shines.
The L1-Ultramaximizer requires the Windows 95 version of Sound Forge and will not work on the regular version designed for Windows 3.x. It comes on a 3 1/2 inch floppy, along with a hardware "dongle" or HASP key. The HASP key has a 25-pin male connector on one side and a 25-pin female connector on the other. Before you can install the software, you have to insert the HASP key into your computer's parallel port. If you use the port for your printer or other device, you disconnect the printer first, install the HASP key and then plug your printer into the end of the HASP key.
The HASP key is a hardware copy-protection device. If you don't insert it before installing the hardware, you won't be able to access the new software. And if you remove it later, you will lose your ability to access the associated software. What do you do if another software also requires a HASP key to run? You just stack them, one after another. Theoretically there is no limit as to how many you can use without adversely affecting your computer's performance. Of course, after a while, you might have to position your computer a few feet from the wall to accommodate all the HASPs!
Once the HASP key has been physically installed on your computer, you're ready to install the L1-Ultramaximizer software. Just insert the disk, and install the program as you would any other new software.
Using the L1-Ultramaximizer as a peak limiter is simplicity itself. There are only two slider controls to adjust. First select a small sample portion of your sound file by clicking the mouse and dragging it along the sound file. You only need a second or two of material.
Then select the Process Menu and the L1-Ultramaximizer submenu. You'll see a screen display similar to Figure One. The only two controls you need worry about for most limiting applications are the Threshold and Output. The Output slider does just what its name implies: It allows you to set the maximum level of the audio signal. While you can set it for the maximum, the Waves manual suggest setting it for slightly less, about .3 dB less than maximum.
The Threshold slider sets the point at which the peak limiter is applied to the signal. If you leave it set at it's maximum point, the software will function just like a regular Normalize processor, evaluating the wave form and increasing the overall level of the wave form so that the peaks are at the maximum level specified.
Virtually all digital audio software provides a Normalize function, and for good reason. Unlike analog recording where you routinely allow the signal peaks to wander above zero dB, digital recording will distort badly if you pass 0 dB by even a fraction. So good recording practice requires staying comfortably below the maximum level. The same good recording practice demands that you later normalize the signal to maximize your signal-to-noise ratio in the system.
The problem with just normalizing a signal, is that many music peaks actually contain little energy and raising the sound wave peaks to 0 dB leaves most of the musical energy well below the maximum level. The L1-Ultramazimizer allows you to raise the overall level of the sound wave with little or no pumping, compression or other undesirable affects usually associated with extreme peak limiting.
To increase the peak limiting, simply lower the threshold level. The manual suggests setting it 4 to 6 dB lower. If you first click on the Preview button, you can then hear the effect of the threshold slider and see it graphically displayed on the Attenuation bar graph. The Preview function will continually loop your selected sample so you will hear the effect of the Threshold slider as you move it up or down.
The Threshold slider determines how much of the sound wave will be affected by the peak limiter. Only that portion of the sound wave that is above the threshold setting will be affected. So if you set the threshold at a minus 6 dB, the software will raise those portions of the sound file that are between the peak and 6 dB below that to peak level.
It's important to pick more than one sample of the sound wave to check it's effect. But once you've applied it a few times, you'll probably find a setting that you'll save and apply to all similar types of program material. When you're satisfied with the amount of peak limiting, simply click on Okay, and the program will ask you if you want to apply it to the whole file or just the sample. After answering that question, the software will apply the peak limiting. On my Pentium 100, the entire process only takes about 70 seconds for a file three minutes in length at 22 KHz resolution. If you don't like the final result, you can undue it just as fast as it was applied.
The L1-Ultramaximizer also has a button for switching between Digital Domain and Analog Domain. The manual recommends leaving it in the Digital Domain mode for most applications. For those times when you may encounter peak overshoot due to a poorly designed DAC or you plan to broadcast the signal without further limiting, the Analog Domain mode will give you brickwall limiting insuring no sample ever exceeds the set ceiling.
Figure 2 shows a portion of a sound file before applying the L1 Ultramaximizer; Figure 3 shows the same sound file after applying approximately 6 dB of peak limiting to the same file. You can see that the file is now significantly higher in overall volume. You'll have to take my word that the sound file sounds great.
That's all there's to it. But as simple as it is to apply, the effect is great. I found that the L1-Ultramaximizer enables me to significantly increase the loudness of a sound file without adversely affecting it's dynamic range. I can easily find just the right combination to increase the overall sound level before cutting in to the natural dynamics of the music. It allows you to significantly increase the perceived level of your audio with little or no negative effects.
Used on a sponsor's promo, it will increase it's impact. Used on music cuts, it will reduce the amount of compression needed from your hardware limiter. In the end, most radio broadcasters want the same thing, a signal that is stronger than their competitors but that is clean and without artificial companding and compression. The Waves L1-Ultramazimizer can help you do that.
But that's not the end of the Ultramaximizer's tools. It also will allow you to requantize your sound files with greater resolution. What does that mean? Quantizing refers to the number of bits you choose to use in working with a file. For high quality recording, most digital audio software programs use 16 bits. Some high quality programs use 20 or 24 bits. Most recorders -- CD or DAT -- are limited to 16 bit resolution.
The L1-Ultramaximizer allows you to work in whatever bit resolution you prefer, and then save it to another lesser bit resolution with a minimal increase in noise by applying specially designed dither algorithms. Where would you want to use this function?
If you record at 24 bit resolution, you'd want to use it if you have to save your file to a conventional CD or DAT in 16-bit form. If you plan on producing a multimedia CD or a Web page with sound bites, you'll probably want to reduce the resulting files to 8 bit resolution and 22 or 11K files. This greatly reduces the amount of file space necessary to store the files.
There are three buttons that effect the Increased Digital Resolution (IDR) function: 1. Quantize; 2. IDR Type and 3. Shaping. Quantize determines the resolution at which the file will be processed: 8, 12, 16, or 20. It's important to note that even though you set the quantize setting to a figure like 8-bits, it will still be a 16-bit file when the processing is done.
Why? All that the L1-Maximizer will have done is set the lower 8 bits of your 16-bit file to zero. So what's the problem? The problem is that it's still a 16-bit file. It will sound like an 8-bit file (More about that in a moment!), but it will be a 16-bit file and if you save it as such, it will take up just as much space as the original 16-bit file. You will have succeeded in lowering the overall sound quality of your sound file while not having gained any reduction in file size.
Then why bother reducing the resolution? You can create a true 8-bit resolution file by selecting that option when you go to save the file. The resulting file will then be a true 8-bit file with the attendant decrease in file size. But it's important to realize that the file as processed by the L1-Ultramaximizer is not a true 8-bit file (or whatever resolution you chose) until and unless you select that option when you save the file.
When you requantize a sound file to a lower resolution, you invariably increase the noise of the sound file. The lower the resolution, the higher the overall noise level. The noise is not unlike the interstation noise you hear on an FM tuner or pink noise. On loud passages, the sound will mask most if not all of it. On quiet passages, it will be quite apparent.
The other two options, IDR Type and Shaping provide you with a means of minimizing this noise through dither and shaping applied to the sound file. The IDR button provides you with three choices: 1. None; 2. Type 1; and 3. Type 2. Type one is described as the purist's choice, adding little or no distortion to the low level portions of the signal.
The Shaping button provides you with four options: 1. None; 2. Moderate; 3. Normal and 4. Ultra. What both the IDR settings and the Shaping options do is to reshape the sound wave to move the noise created by requantization to a portion of the audio spectrum in which the human ear is less sensitive. In other words, it doesn't reduce the noise, but it reshapes it in such a form that it is less apparent and/or less noticeable to the human ear.
The software is designed to provide the greatest effect with 48K 16 bit sound files. But it also can improve the sound quality of sound files requantized at other resolutions. The preview function is extremely helpful in determining which combination of settings to apply to a particular file. You can quickly try all combinations and decide which one is best for the current file. It's a matter of individual taste and application.
Since the noise introduced by lowering the resolution of a sound file is essentially a low level noise, adjusting the peak limiter function to a higher level of limiting can mask more of the resultant noise. I applied the IDR functions to both 48K and 22K files, and found that it does improve the quality of sound files requantized to a lower resolution.
Waves has an extremely well written manual that not only walks you through all of the functions of the L1-Ultramaximizer, but gives you an in-depth background on the theory and issues affecting peak limiting and quantization. Their web site contains even more information on using the L1-Ultramaximizer and other Waves products.
K.S.Waves, Ltd., is a software company specializing in signal processing and user interfaces for the professional audio and multimedia markets. Waves began operation in 1988 with offices in Tel Aviv, Israel, and Knoxville, Tennessee. Founders Meir Shashua and Gilad Keren developed DSP mathematical algorithms for audio processing on PC and Mac personal computers.
One of their initial products was the Q10 Paragraphic EQ software tool for Digidesign Inc.'s digital audio workstation. Since then they have continued to develop a host of digital audio plug-in's for use on a variety of digital audio software for both the PC and the MAC.
The L1-Ultramaximizer is produced by K.S.Waves as a plug-in for Sonic Foundry's Sound Forge. It requires Windows 95 and has a list price of $595. Sonic Foundry has been offering it at a special introductory price of $450.
As far as I'm concerned, this is an excellent product that moves Sonic Foundry's Sound Forge into world class digital audio standing. If you plan on mastering CD's or just want to give your sponsors the hottest spot around, Waves L1-Ultramaximizer will give you an edge up on your competition. Until they get their own copy.
-- The End --
Sonic Foundry, 100 South Baldwin, Suite 204, Madison, WI 53703, (800) 577-6642, Internet: firstname.lastname@example.org, Home Page: http://www.sfoundry.com.
US Waves, 4302 Papermill Road, Knoxville, TN 37909 (423) 588-9307 Email: email@example.com.
Read Burgan is a free lance writer and a former public radio station manager who can be reached at (906) 296-0652 or through e-mail at firstname.lastname@example.org.