SEARCH AIG







 

  ETF Listening Room Analysis Software

      Date posted: April 9, 1999


AN ETF WATERFALL PLOT
Sugg. Retail: US$199.95
Manufacturer: ETF Acoustic Ltd.
http://www.etfacoustic.com

Room for Improvement

by Gordon Brockhouse

      For audiophiles, listening rooms are a bit like the weather. We talk about them, but few of us do much about them. We might experiment with speaker placement and listening-chair position, and some ambitious souls try to tame room problems with products like ASC Tube Traps. But not many of us go beyond that. That’s too bad, because as we all know, the listening environment is second only to the loudspeakers in its influence on sound quality.

      Unlike the weather, we can do something about room behavior - if we approach it rigorously. That means identifying room problems, their effects and possible solutions. You can’t do that by ear alone. Even simple experiments with speaker placement and seating position are difficult. After you’ve spent a quarter-hour hefting your couch and speakers about, it’s hard to be sure whether “after” sounds better than “before.” And it’s impossible to know if there are room problems in play. What we need is a tool to complement our ears.

      I’ve spent the past few days checking out just such a tool: ETF 4.0, a software package developed and sold by ETF Acoustic of Oshawa, Ont. “ETF” stands for “Energy Time Frequency,” and I infer from the numerical designation that this is the fourth version of the package. ETF says the program should run on any Pentium-based PC running Windows 95, Windows 98 or Windows NT.

      I’ll state my opinion right up front: ETF 4.0 is an excellent product, well worth its US$199.95 price tag. (While ETF is a Canadian company, all its transactions are in U.S. dollars.) Anyone who’s willing to spend some time with the product, and has enough flexibility to implement measures such as diffuser panels or bass traps, should find that this investment yields very significant sonic dividends. Actually, you don’t have to spend a nickel to obtain one of the ETF 4.0’s great benefits: an education. Together, the superb manual and extensive help files provide a mini-course on room acoustics and their effect on sound quality. A trial version of ETF 4.0, available free of charge from ETF’s Web site (http://www.etfacoustic.com), includes this material, plus a “crippled” version of the ETF software.

      Documentation is often the weak point of software from small companies, but that’s not the case here. A model of clarity, the 30-page manual begins with a primer on listening room acoustics, then proceeds to a description of the ETF system: the measurements it takes, and how to analyze them. Along the way, there are suggestions for taming room problems. The help files elaborate on this subject matter, as well as providing detailed guidance to using the software and interpreting results.

Common Room Problems

      The main problems affecting listening rooms under 10,000 cubic feet (i.e. almost all domestic listening rooms) are room resonances and early reflections. Room resonances cause audible problems at lower frequencies, where sound is non-directional, and behaves like a wave. When room resonances are severe, the result is uneven bass response and a “boomy” character. This is a common problem in domestic listening environments. Room boundaries near the speaker can also cause uneven bass response.Early reflections cause audible problems in frequencies above 300-400 Hz.

      As frequency increases, sound becomes increasingly directional; it behaves like a ray of light. When two acoustic events occur within a very short time of one another (less than 15 to 20 milliseconds), we perceive the two events as one. This is known as the precedence or Haas effect. Reflections which occur within this window cause audible interference; some frequencies will be reinforced, others cancelled. This is called “comb filtering” because of the jagged appearance of a frequency-response graph showing the effect, which is to change the perceived frequency response and to blur ambient information present in the recording itself. Typically, ceiling and floor reflections are the worst sources of early reflections, but reflections from the side walls can also cause audible problems. As long as later reflections are smooth, and decay gracefully, they’re desirable. But in some rooms the response may vary widely as the sound decays, which can make the room sound harsh.

Collecting Data

      ETF 4.0 lets you pinpoint all these problems. The first step is to collect data. You do this by playing a series of test tones and recording them onto your computer. The software analyzes the recorded data, and produces a series of informative graphs. There are several ways to collect data. What I did was hook the line output of my computer’s sound card to the line input of my preamp so that the test tones would play through my PSB Stratus Gold loudspeakers. As ETF recommends, I tested one channel at a time. Rather than connecting a microphone directly to the sound card’s mic input, ETF suggests using a microphone preamplifier, and taking the signal into the sound card’s line-level input. The company sells a calibrated mic and matching preamp for US$199.95. As an alternative, ETF suggests a Radio Shack sound level meter, which has a preamplified output. If your computer has a “full-duplex” sound card (i.e. if it can record and play at the same time), you should be able to play a WAV file on the hard disc containing the test signals while recording the results of the tests as another WAV file. I couldn’t do this with the Ensoniq sound card supplied with my Hewlett-Packard Pavilion system, so I substituted a SoundBlaster AWE64 Gold card, which worked fine.

      After putting the mic in the listening position and hooking everything up, I launched the ETF software, then chose the “New” option under the File menu, and selected “Play/record test signal.” Up came a box with a record/play button. When I pushed it, a series of sweeps, chirps and whistles emanated from my left speaker, and a peak-level reading of the recorded signal appeared on the screen. A screen appeared, telling me that the recording process was finished. I clicked “Next,” and the computer went away to think for a while. A half-minute later, a graph showing the decay of the sound in the room (an energy-time curve) appeared on my screen. My HP system has a 333 MHz Pentium II processor and 64 MB of memory. With less powerful systems, the analysis can take several minutes. I saved the resulting file, then repeated the process through the right channel. I did this using both the ETF calibrated mic and the mic in my Radio Shack sound level meter. (The curves I got using the meter were somewhat different from the ETF mic and preamp; but generally, recordings made with the meter led to the same overall conclusions about my room’s behavior.)

      Besides the energy-time curve, ETF shows several different frequency-response curves and harmonic-distortion graphs. Analysis of this information can tell you a lot about your listening room. Before discussing this, it’s worth noting various options for gathering data. ETF 4.0 includes a test-signal file that you can write to CD-R if your computer has a CD burner; this will play on a regular CD player. If you don’t have a CD burner, ETF will ship an audio CD containing the test signals. To use it, you place the mic in the listening position, start recording and play the test signal through a regular CD player. ETF says this method is preferred for audiophiles over playing test signals from the computer’s sound card. Most consumer sound cards have uneven frequency response and that will skew results, the company says. However, it also provides a calibration function to compensate for sound-card non-linearity. In addition, this method will work with notebook computers, many of which have half-duplex sound processors.

      When playing test signals over a CD player, you can record results from within ETF or with the Windows sound recorder (a Windows utility for making digital recordings through the sound card; with ETF be sure to set your card for 44.1 kHz 16-bit mono recording). Or you could record onto audio tape (preferably a high-fidelity medium such as DAT), then transfer the recording to your computer later on using the Windows sound recorder. You then open the resulting WAV file with ETF and proceed with your analysis.


FIGURE #1

FIGURE #2

 

Figure #1. The energy-time curve shows how the output from my right speaker decays over 50 milliseconds.

Figure #2. This frequency-response curve has 3-millisecond gating which excludes most room effects.

FIGURE #3
   
Figure #3. This frequency-response curve shows the effects of reflections occurring in the first 10 milliseconds.
   


Interpreting the Results

      The energy-time curve will tell you whether early reflections are a problem in your setup. The curve maps the rate at which four different areas of the audible spectrum decay. Ideally, what you’ll see is a peak at about 10 msec, then little or nothing for the next 15 msec, then later reflections 6 to 12 dB below the level of the direct sound, which gradually fade away. As you’ll see from the energy time curve for my right channel (figure 1), there’s certainly a lot happening in the 20 msec following the direct sound, particularly between 500 and 1000 Hz. You can see the effects in the frequency response graphs. Figure 2 shows frequency response between 500 Hz and 10 kHz with 3-msec gating, meaning that the system excludes sounds arriving at the microphone more than 3 msec after the direct sound. A short gating period will exclude most room effects, but may also miss resonance problems in the speaker itself. Figures 3 and 4 show frequency response between 500 Hz and 10 kHz with 10- and 20-msec gating respectively. You can see the massive effects of the early reflections on frequency response. Fortunately, our ear-brain can sort through much of this confusion, so that the system doesn’t sound nearly as bad as these curves look! Still, if we can find a way to limit these reflections, the system should definitely sound smoother and image more precisely.

FIGURE #4

FIGURE #5

Figure #4. This frequency-response curve shows the effects of reflections occurring in the first twenty milliseconds.

Figure #5. The regular peaks and dips in this linear frequency-response curve help identify the source of early reflections.

FIGURE #6
   
Figure #6. This 3D curve shows how frequency response changes over a 120 millisecond period.
   

      Switching from the familiar logarithmic view of frequency response to a linear view (figure 5) helps us zero in on the problem. We see a recurring series of peaks and dips. They’re fairly regularly spaced - about 300 Hz apart. Frequency is the inverse of time, which means that the reflected sound causing this interference is arriving 1/300 second, or 3.3 msec, after the direct sound. Sound travels about one foot per millisecond, so the path the reflection is taking must be about 3.3 feet longer than the direct path. In my room, it looks like a ceiling reflection is causing this particular problem.

      Another set of ETF charts will show the effects of both early and later reflections. The 3D graph (figure 6) shows frequency response over a 120-msec period. This particular graph is quite hard to interpret, but the “slice” button allows you to look at response in discrete time segments. Figure 7 shows a slice at 13.3 msec, 3.3 msec after the initial sound. You can see the effect of the ceiling bounce quite clearly. Figures 8 and 9 show the response at 35 and 36.7 msec. The curves are anything but smooth and are also quite different, indicating problems with late reflections.

FIGURE #7

FIGURE #8

Figure #7. This time slice shows how a ceiling reflection affects response.

Figure #8. A subsequent time slice shows how later reflections affect response.

FIGURE #9
   
Figure #9. A subsequent time slice shows how later reflections affect response.
   

      You use a similar process to isolate low-frequency resonances. The low-frequency curve (figure 10) shows rather uneven bass response, while a 3D graph of low-frequency response over a 1800-msec period (figure 11) show some resonances (you see peaks in the early curves, and sustained output at some frequencies in later curves. You can isolate the resonant frequencies by looking at later time slices. Figure 12 shows a major resonance around 37 Hz and smaller peaks around 52, 74 and 100 Hz.

FIGURE #10

FIGURE #11

Figure #10. This curve shows right-channel bass response in my room.

Figure #11. In this 3D curve we see some resonances that persist after other sounds have died away.

FIGURE #12
   
Figure #12. With this time slice we can indentify where these resonances occur.
   

Fixing the Problems

      So what do we do about these problems? The first line of attack is the usual one: experiment with speaker and listening-chair position. Low-frequency resonances are caused by room modes (interaction of low-frequency output with room dimensions) and room boundaries (reinforcement of bass output by room surfaces). As we all know, moving speakers and listening chair can go a long way toward taming uneven bass response. Reorienting the speaker or changing its location can also reduce the effects of early side-wall reflections. With ETF, you can do before-and-after comparisons to determine whether the new position is an improvement over the old one, and to find out if it’s introducing new problems. You’ll find this more reliable and revealing than relying on your ears alone.

      From there we move to room treatments. Generally, floor and ceiling reflections can’t be addressed with a simple shift in speaker or listening-chair position. Regardless of the position, there will always be a strong ceiling and floor bounce. You can carpet the floor and treat the ceiling with sound-absorbing foam. My family room has a carpeted floor but a very hard plaster ceiling, so it’s no wonder its effects are so significant. I’ll have to live with the problem; a foam treatment for the family-room ceiling will be an impossible sell in our household.

      ETF suggests using diffusion panels to tame problematic reflections that remain after you’ve finished moving speakers and listening chair. The manual describes a simple means of locating sources of early reflections. It’s called “the mirror trick.” While you sit in your listening chair, have someone move a mirror along walls, ceilings and large objects in front of you. Any mirror location where you can see the loudspeaker is a source of early reflections. If those reflections are problematic, you can tame them by putting a diffusion panel there. For taming bass problems that remain after your back’s given out, you can try building some bass traps, and putting them in the back corners. The trap acts as a bass sponge, soaking up excess low-frequency energy. ETF’s documentation includes instructions for building a bass trap called a “Helmoltz radiator,” a tall ducted cylinder. There are also instructions for building diffusion panels. The package includes “Device Designer,” a software utility that calculates the effects of different designs, so that you can build something that addresses the problems in your setup.

      There’s more to ETF than I can convey in the space at hand. It seems to be a very solid product. In several days of use, I did not experience any crashes or other significant hiccups. About the only criticism I can make is that you can’t really use the program at screen resolutions under 800×600, which may rule out the product for some older and low-end notebook PCs. As I’ve already mentioned the documentation is first-rate, and support seems to be good as well. In the trial version, you can’t access some of the more useful functions. You can’t save, so you can’t do before-and-after comparisons. The linear frequency-response graphs are disabled, which makes it more difficult to isolate early reflections. So are the low-frequency response graphs, which means you can’t use the software to track down resonance problems. But you can view energy-time and mid/high frequency-response curves, as well as distortion curves.

      Equally important, you can get a feel for the product, and avail yourself of its educational value. You can upgrade to the full version by contacting ETF and paying the US$199.95 license fee. The company will give you a code that unlocks all the crippled functions. ETF will also send you the software and the audio CD containing test signals if you pay a shipping fee in addition to the license fee. For serious audiophiles, as well as audio and home-theatre specialists, I can’t think of a C$300 accessory that will deliver as much value as this product does. It won’t do much about the weather, but it’ll sure help you do a lot about your room.

Gordon Brockhouse

Share This on Your Favourite Social Networking Site:
  • Facebook
  • Technorati
  • StumbleUpon
  • Digg
  • del.icio.us
email the author

Related Reviews:

No related posts

Comment On this Post

  Outside the Speakers

    Random Thoughts on the Music Mask
    NPR on Whether Audiophiles Still Exist
    Audiophile Grade Mics?
    CDs Sales Die, LP Sales Fly
    Some High End 'Phones from CES

Got a tip? news@audio-ideas.com
  AIG Reviews
Loudspeakers | HT Loudspeakers | Power Amplifiers | Preamplifiers | Integrated Amplifiers | Receivers | Surround Processors | Digital Sources | CD Players | DVD Players | Network Music Players | SACD Players | Analog | Cables | Subwoofers | Portable Audio | Pro Audio | Headphones | Systems | Tuners | Video | Vintage Gear | HD Video | Blu-Ray | Online Archives

Complete List of Online Reviews
  Columns

     Audio Ideas (Andrew Marshall)
     Ox Box (Bob Oxley)
     Hy End (Hy Sarick)
     Bain's Blog (John Edward Bain)

Interested in Writing about Hi-Fi and Home Theater for AIG? Click to Email a Writing Sample

  Most Popular Today
  1. Behringer 1002B Portable/Studio Mixer - A Remarkable Recording Tool, Especially At The Price!
  2. Paradigm Monitor Series 7 Sub 10 Subwoofer - Extended, Clean Bass From A Tiny Box!
  3. Audio Ideas: CBC Radio Two - Intelligent Music Selection Becomes A Sausage Factory
  4. A Tale of Two Headphones from Canada by NAD VISO & Paradigm!
  5. Soundsmith Aida Ebony Phono Cartridge: Looking For Your Last Best Cartridge? This Could Be It, At A Decent Price!
  6. Bryston B135 SST² Integrated Amplifier/DAC
  7. Audio Ideas: Open Reel Wow and Flutter? It May Be The Pinch Roller!
  8. Audioquest DragonFly USB DAC
  9. Origin Live Calypso Turntable and Encounter Tonearm
  10. AUDIO IDEAS: Not-So-Audio-Related Thoughts About Clouds, Passwords & Controlling Remotely!
  Most Popular Overall
  • Recent Comments: