How does the Wavelet Bohmer approach to room correction differ from others?
Dudleston: It’s all in the timing. The sound of music is created by varying air pressure over time. To weigh just how important time is to music, consider the units of measure for frequency or pitch (Hz) are cycles per second. So it must be well understood that precise timing in the arrival of energy is fundamental to accurate musical reproduction. Vary a turntable’s platter rotations/time, and the pitch changes. Change the time between notes and the tempo changes.
Anyone measuring a loudspeaker with conventional methods in a listening room quickly learns that obtaining a measurement that correlates to how the loudspeaker sounds is virtually impossible. The measurements vary greatly with the location of the microphone. Measuring the speaker at a number of locations about the room to create an average response may generate a generalized plot, but will not incorporate precise time domain data. One cannot logically omit such vital timing information when correcting a loudspeaker to a room.
Ironically, it is in the time domain that we find the real solution to the problem. It is the combining of multiple arrivals that cause problematic response dips and peaks. The room is dominating the decay of power. As a result the anechoic “flat” speaker may exhibit boominess or muddiness, and transient hangover. Attempting to equalize the response without regard to the time domain will actually compound the response errors at other points in the room while creating new time domain errors at the listener position.
The key differences are:
- The algorithm relies on psychoacoustic weighting (how we hear) and how the sound sequentially arrives to us.
- The applied correction is not sensitive to position. Transient behavior is improved on- and off-axis.
- Microphone placement is similar to how you record an instrument, at a distance of about four feet from the speaker on a direct path to the listener. The objective is to measure how the room is loading the speaker by measuring sound pressure over an event window up to several hundred milliseconds at each frequency.
- The measurements are readily captured and exported to a dedicated website where a super-computer performs many thousands of iterative calculations to determine the optimal correction solution.
- After the correction algorithm coefficients are rendered for each speaker and automatically downloaded into the processor, the processor can apply the correction in real-time over a full 40ms window.
- Psychoacoustic weighting is emphasized in the calculations. The time domain is not compromised to fill minor dips in the response curve. The hearing mechanism is given adequate, accurate cues and left to perform its tasks. Aligning the energy in time is what smooths the response curve, not power equalization.
Cordesman Assessment of Impact on Sound: First, I should note that setup was exceptionally quick and easy, and not tied to a particular listening position. I’ve had quite a bit of grief getting good results with some other systems that required multiple setups to test different microphone locations. I would, however, try out the Legacy V with the room-correction feature turned off to first find the best speaker location without it, and then apply the DSP to that setup.
Also, if you do audition this speaker be aware that switching the room-correction feature has the apparent effect of slightly reducing volume. You can’t just A/B.
That said, I’ve already stressed many of the advantages of the room-correction features, and I should emphasize that I use it with the Legacy V speakers placed in areas in my room with minimal resonant effects. In some ways, this means the benefits of room correction are lower than would be the case in most rooms, and certainly lower than in problem rooms—few of us can build a house or listening room around our systems.
In most setups, it is clear that the overall mix of room-correction features with good acoustic recordings will not perform some sonic miracle. They will, however, provide an improvement in the clarity of the music over a wide range of frequencies, an increase in soundstage depth and imaging, and an audibly clearer lower midrange and mid-to-upper bass.
This is particularly apparent with the bass viol, cello, piano, lower woodwinds, guitar, and natural-sounding percussion recordings, and also in clearing up vocals, naturally staged operas, and choral music. It is also far more apparent with really good recordings. No system can make up for what is not there in the first place. The Wavelet can’t fix what isn’t miked naturally for location and realistic soundstage, although the sound of crowd noise and applause sometimes does become more natural.
I should stress that the sound of the best conventional speakers in a really good system with a really good setup is still very competitive in overall quality. Moreover, every audiophile has his own preferred set of nuances and sound characteristics, and no system can be all things to all audiophiles. At the same time, the room-correction features have real advantages that do make both the speaker location and listening position far less critical.
What kinds of frequency and timing correction are being applied, and what are their limits relative to making everything flat?
Dudleston: We address the sequence of reflections from boundaries that occur in the first 40ms. The correction works almost entirely in the time domain. The time-corrected response is always smoother but never ruler-flat. The bass will be deeper with faster decay. Vocals will be full, but not chesty or weak from the floor bump and dip that nearly always occurs.
Sonically this translates to a more natural transient. If an actual drum is struck, the initial sharp transient pulse will be felt as well as heard. Conventional speakers without controlled directivity and proper room correction can smear the event in time, compromising tautness. This is readily evidenced by defeating the room correction, even on the highly controlled V system.
The loudspeaker and crossover are matched carefully to the target function before the speaker leaves the factory. Each section of the speaker is already compensated by the Wavelet in time before room correction is applied. Frequency adjustments are gently applied to each of the drivers with respect to time domain. The speaker is minimum-phase in behavior and the response quite smooth. While we have the luxury of already knowing precisely what the speaker is delivering acoustically into the room from the floor up, even at what radiation angle it is being launched, the room still gets the last word. As room boundaries are closer than one-quarter wavelength, the radiation efficiency actually increases with initial reflections. As the frequency rises, cancellations begin to occur. Corrections for these phenomena are not accomplished by boosting or cutting the amplitude, but by literally pushing the late energy back to the original event. Both the significance and audibility of these corrections decreases progressively above 500Hz in most listening rooms.
We have learned that notch filtering of resonances is not a solution. When a dip is boosted in the power response, so is the reflection that caused the dip in the first place. Amplifier power is wasted, the speaker works harder, and the time domain is corrupted. Many of the nulls and buildups we experience are not from resonances at all, but instead from reflections interfering with the direct path to the listener. Remember that multiple reflections are required to create a resonance, so the time required for them to generate is significantly longer. Resonance and reflection behavior is quite different and should be treated accordingly. Automated equalization of the power response is treating a symptom, not the cause.
A chief problem in correction systems lies in the unevenness of sound throughout the room. It originates because of the way correction systems treat the different resonant phenomena within the room. They tend to disregard the origin of the resonance, treating them all in an equal manner. This is a mistake because resonances with different origins have radically different behavior.