Music in the Round #23 Page 2

Basically, the installer enters general information about the system, including reference data about the components and room dimensions, and exactly which channel signals are connected to each of the SE's eight inputs/outputs. This is because you can use these channels any way you need to: just two for stereo, or eight for a 5.1-channel system and a two-channel system, or eight for a 6.2 system—or even eight for four stereo systems in four different rooms. Clearly, you've got to tell the SE what's connected to what. My configuration was pretty basic: a 5.1 system connected to input/output jacks 1–6. SE channels 1–4 are of higher resolution than channels 5–8, so I made sure my front three channels were connected to them.

You connect your PC to the SE via the latter's USB port, and the calibrating microphone via its separate preamp to Input 1, in place of the normal input cable—but only when the installation software prompts you to do so. The software then generates 10 chirps for each channel in use. The first set of chirps is generated and measured with the microphone in the preferred listening spot, and those also determine the level trims, delays, and crossovers. After this, you're prompted to reposition the mike and repeat the measurements. Each time, you have the option to delete and repeat a measurement—if, for example, it was subject to excessive noise, such as a sneeze or a passing car.

The goal of these additional measurements—up to 31 in addition to the first—is to determine a set of correction filters that will accommodate more than just a single listener with his head clamped in the sweet spot. Also, the nulls and nodes of room modes are often steep and highly localized with respect to room boundaries, and you can't be sure that any given measurement site is truly representative. I did the procedure several times, each time placing the mike in six to twelve places that differed in lateral and vertical position.

After the chirp tests, the PC churns through its calculations and delivers its Detection Results. These include, for each channel: speaker distance (for delay settings), speaker polarity (check your connections), level trim (to equalize volume), and recommended crossover frequency (with override option). All the results were very close to those I got from my NHT Controller's auto setup. Note that all of these Detection Results are not set in the Sound Equalizer, but must be transferred by the person performing the installation to your A/V receiver or preamplifier-processor. Note, too, that you must accept or set the MultEQ Pro software for the chosen bass-management frequencies, because the EQ filters, which are yet to be calculated, will incorporate that information.

Next, you must choose one of four target frequency responses that range from flat to 20kHz with varying degrees of HF rolloff. I chose Flat, though for my small room Audyssey recommends a slight rolloff. MultEQ Pro then calculates the EQ filters for each channel and prompts you to transfer them to the SE. At this point, you can view before-and-after response graphs for each channel while you still have the option to make more measurements. Unfortunately, the graphs are uncalibrated (at what frequency is that bump?), say nothing about spectral decay, and are highly smoothed, especially at high frequencies (footnote 1). They may be adequate to tell you what you've done, but you can toggle MultEQ Pro on or off from the PC to subjectively assess whether you've accomplished your goals, before permanently saving them on the SE.

What I heard: The change between MultEQ Pro on and off was readily apparent but not huge. Most noticeable was what I call a trimmer waistline—the loss of lumps or emphasis in the mid- to upper bass, which made the overall balance sound clearer and cleaner. Imaging was greatly improved—individual voices and instruments were more precisely locatable, but with no deterioration in the integrity vague of larger forces. Finally, the low bass was more tight and detailed, and less noticeably separate from the rest of the audioband.

Could this be verified? Measurements were performed under two different situations. My neighbor, Ethan Winer, of RealTraps, did the first set with Acoustisoft's ETF 5. (I wrote about the earlier ETF 4 in July 1998.) (Ethan will no doubt offer his data and insights on his own website, after this issue of Stereophile appears in print.) For this, my first pass with Audyssey MultEQ Pro, we removed the Velodyne SMS-1 Sub EQ and a pair of RealTraps TriCorner room treatments from the setup, and would have removed more had it been convenient. Also, at the time, I didn't yet know to transfer the trim settings to the pre-pro I am using, and all speakers were set, as suggested by MultEQ Pro, as Large. Ethan's measurements show small corrections in overall frequency response and even smaller ones in decay times, particularly at low frequencies. Despite this, we did hear a slight improvement in the sound, along the lines of what I've described above.

I made the second set of measurements with both Goldline's TEF system and ETF 5, but under different conditions. First, I reinstated the TriCorners, which I've found make a huge subjective difference in the midbass. Second, I remeasured, recalculated, and scrupulously transferred all trims to the pre-pro using a wider, more varied (if less geometric) range of microphone positions. Third, I set the surround speakers to Small, with an 80Hz crossover point.

Did it sound better? I can't say the change was blatant, but over the course of weeks it became increasingly apparent that the system without MultiEQ Pro sounded simply inferior, in terms of the clarity of sounds from the low bass to the upper treble, and in terms of imaging all around the compass. My measurements confirmed this, especially in the midbass, though I offer here only one or two examples of the improvement (figs.1 and 2, right channel before and after, measured with ETF 5; figs.3 and 4, subwoofer before and after). What was surprising was that MultEQ Pro made a bigger improvement with this improved setup. This suggests that MultEQ Pro is not a panacea, but requires an already well-configured system in a reasonably good acoustic environment—and Audyssey says as much. To correct the sound of a room whose surfaces consisted entirely of tiles and windows would be asking too much.

Fig.1 Right-channel speaker, in-room cumulative spectral decay plot, 20–200Hz, without equalization, measured with ETF 5. (10dB/vertical div., 800ms time window)

Fig.2 Right-channel speaker, in-room cumulative spectral decay plot, 20–200Hz, with Audyssey equalization, measured with ETF 5. (10dB/vertical div., 800ms time window)

Fig.3 Subwoofer, in-room cumulative spectral decay plot, 20–200Hz, without equalization, measured with ETF 5. (10dB/vertical div., 800ms time window)

Fig.4 Subwoofer, in-room cumulative spectral decay plot, 20–200Hz, with Audyssey equalization, measured with ETF 5. (10dB/vertical div., 800ms time window)

Bottom line: The Audyssey Sound Equalizer with MultEQ Pro genuinely enhanced the sound of my system. Once it was installed, I could simply leave it alone and enjoy a more transparent and coherent soundstage. The SE is obviously designed for the lover of music and home theater who is not interested in toys or technology, but in the best possible sound.

Even with its very audible improvements and audible deficiencies, the Audyssey SE won't make the inveterate audiophile or technophile happy. It cannot store multiple filter sets optimized for different styles of music or listening positions, and it requires a professional installer initially and for any system updating. Still, any serious listener who can't have a purpose-engineered listening and/or viewing room must hear what the Audyssey Sound Equalizer can do.

Footnote 1: If your installer uploads your system's original data to, Audyssey will send you Before and After charts of each channel's frequency response. The charts they sent me were beautiful but, I feel, optimistic, and inconsistent with what Ethan and I measured on-site.