[Continued from Part Two. . . .]
Flat Frequency Resonse and Conquering the Sound of the Listening Room
As far as I can see, Gradient's "The Absolute Listening Test" conclusively establishes that under anechoic conditions all that really matters in terms of the realism of a loudspeaker being able to accurately reproduce its input signal is flat on-axis frequency response. The problem is, this does not hold true once you listen to speakers in any sort of ordinary listening room which is not anechoic. The problem is knowing how to design a speaker so that it can accurately reproduce its input in a real listening room given the corrective tools we can apply. Those corrective tools include placement of the listener and speakers, passive and electronic equalization, and acoustical room treatments.
The passive acoustic tools include bass traps and absorptive and diffusive panels. These can be used to create a Reflection Free Zone (RFZ) and eliminate at least the most troublesome bass resonances and all the troubling reflections. RPG, for example, has a full line of products. Digital EQ can eliminate any other bass modal problems. While these methods are both costly and ugly, they do exist for the truly dedicated and well-heeled. Even my Sonex is a huge step in the right direction, and doing a room with 3-inch Sonex should cost $1,000 or less. The full RPG treatment plus bass traps could easily run $30,000.
The fact that many commercial recordings sound weird played back outdoors or in a RFZ room is not because the RFZ is not doing its job. If the commercial producers were to engineer recordings to be "wet" enough to play back optimally in a RFZ room, many listeners with more lively home acoustics would find such recordings overly reverberant and "swimmy," or "too distant" sounding. Recordings are intentionally made too close up, with not enough hall sound. Beside the mike perspective effects discussed above, commercial recording producers also assume that most listeners are not about to create a RFZ home listening environment. Commercial recordings of classical music do poorly enough sales-wise as is, without further hampering sales by catering to the hardcore audiophile perspective. Thus, overly dry recordings are the intentional norm since most listening rooms are assumed to add room sound, albeit the wrong kind. This is the second reason why I assume that most commercial recordings will never offer the engineer’s best shot at achieving a “you are there” perspective in the home listening room: any such best shot would have to assume that the recording has captured all the relevant recording venue ambience from the proper perspective and that the home listening room effectively has no acoustics of its own.
Conquering Floor Bounce
Floor bounce is, by definition, the product of a reflection of sound from the floor between the listener and the loudspeakers. The result is usually a suck out in the frequency response in the orchestral power or warmth range between 100 and 250 Hz, the affected frequencies depending on how far above the floor the woofer is mounted. A reflection free zone is, by definition an area in the listening room where early reflections, such as floor bounce, are not audible because they are absorbed or diffused. Thus, an RFZ in the listening room will take care of any floor bounce problem. End of story. Again, it may be ugly, expensive, and inconvenient for listening room occupants in terms of what needs to be done to the listening room to deal with the floor bounce, but it definitely CAN be done with passive acoustical tools, and can be done for the bounce from any number of channels, not just two. Studios physically deflect the audibility of the floor bounce from the front channels by positioning the mixing desk between the listener and the specular (mirror-like) reflection point on the floor. Very thick (say, a foot) floor padding will also work down in the 100 – 200 Hz region of this problem.
Proper Speaker Radiation Patterns
Equalization, even sophisticated digital signal processing, can only efficiently correct the direct sound coming from the loudspeakers, not the sound reflected off room surfaces. This may not be obvious, but it is true. It is one of the flies in the ointment of the efficacy of all attempts to electronically equalize system response using either analog or digital electronics. With most speakers, broadband DSP equalization may help, but leaves some listeners, me included, somewhat dissatisfied. The inability to correct the indirect or room sound is one reason why.
Equalization works best for low frequencies, those below 300 Hz or so. Even with low frequencies, I find the result problematic, with measured flat response in this region usually sounding somewhat anemic. I have to add bass level to restore realism and if I move my head a few inches the bass is not subjectively flat anymore.
Recently, REG of TAS has come to the conclusion that, for maximum realism and reciprocity to the recording, home speakers with multiple drivers should have wide (as in at least wide enough to surround a 12" bass driver) front baffles to avoid baffle-step effects (sudden changes in directionality of radiation) in the midrange. Home loudspeakers should also have constant directivity up to about 4 kHz, above which frequency directivity should increase. REG reasons that only such a radiation pattern can be made subjectively flat using even digital equalization for the lows and common room treatment to absorb high frequency reflections. Only for such speakers can the midrange room reflections be easily minimized in their audible effects via passive acoustical treatments and good speaker and listener placement. REG began his exploration of this topic here, especially beginning at page 16 with the part labeled "The Myth of Wide Dispersion and What Actually Makes Speakers Work in Rooms."
As is typical, REG's conclusions run counter to current audiophile speaker design practice, which either emphasizes narrow front baffles and wide, uniform dispersion to very high frequencies, or which goes to the opposite extreme, producing ever-increasing directivity starting at a very low frequency. REG finds the typical narrow-baffle speakers to be too bright sounding in upper mids and highs even when measuring flat on axis. I agree. He finds those speakers which have decreasing directivity from a very low frequency on up to sound too lifeless.
According to REG, one example of a currently available, reasonably priced (less than $3,000 a pair) and reasonably sized (about two cubic feet—a large "bookshelf" sized) speaker which should be able to work exceptionally well in a home listening room is the JBL Pro LSR 6332. REG has noted that pro-audio folks are well aware of the need for a constant directivity index in the midrange and thus it is not surprising that some speakers aimed at the pro-audio market, like this JBL, are designed to deliver this result. Note in the graphs how the on-axis directivity index is practically constant up to 4 kHz, how well the directivity index of the direct sound follows that of the first reflections, and how closely the frequency response at 30 degrees off axis follows the response on axis, with the off-axis sound falling off in response above 4 kHz.
One might wonder about why other types of speaker designs cannot solve the reproduction problems when used in real rooms. Here are my observations:
Flat Frequency Resonse and Conquering the Sound of the Listening Room
As far as I can see, Gradient's "The Absolute Listening Test" conclusively establishes that under anechoic conditions all that really matters in terms of the realism of a loudspeaker being able to accurately reproduce its input signal is flat on-axis frequency response. The problem is, this does not hold true once you listen to speakers in any sort of ordinary listening room which is not anechoic. The problem is knowing how to design a speaker so that it can accurately reproduce its input in a real listening room given the corrective tools we can apply. Those corrective tools include placement of the listener and speakers, passive and electronic equalization, and acoustical room treatments.
The passive acoustic tools include bass traps and absorptive and diffusive panels. These can be used to create a Reflection Free Zone (RFZ) and eliminate at least the most troublesome bass resonances and all the troubling reflections. RPG, for example, has a full line of products. Digital EQ can eliminate any other bass modal problems. While these methods are both costly and ugly, they do exist for the truly dedicated and well-heeled. Even my Sonex is a huge step in the right direction, and doing a room with 3-inch Sonex should cost $1,000 or less. The full RPG treatment plus bass traps could easily run $30,000.
The fact that many commercial recordings sound weird played back outdoors or in a RFZ room is not because the RFZ is not doing its job. If the commercial producers were to engineer recordings to be "wet" enough to play back optimally in a RFZ room, many listeners with more lively home acoustics would find such recordings overly reverberant and "swimmy," or "too distant" sounding. Recordings are intentionally made too close up, with not enough hall sound. Beside the mike perspective effects discussed above, commercial recording producers also assume that most listeners are not about to create a RFZ home listening environment. Commercial recordings of classical music do poorly enough sales-wise as is, without further hampering sales by catering to the hardcore audiophile perspective. Thus, overly dry recordings are the intentional norm since most listening rooms are assumed to add room sound, albeit the wrong kind. This is the second reason why I assume that most commercial recordings will never offer the engineer’s best shot at achieving a “you are there” perspective in the home listening room: any such best shot would have to assume that the recording has captured all the relevant recording venue ambience from the proper perspective and that the home listening room effectively has no acoustics of its own.
Conquering Floor Bounce
Floor bounce is, by definition, the product of a reflection of sound from the floor between the listener and the loudspeakers. The result is usually a suck out in the frequency response in the orchestral power or warmth range between 100 and 250 Hz, the affected frequencies depending on how far above the floor the woofer is mounted. A reflection free zone is, by definition an area in the listening room where early reflections, such as floor bounce, are not audible because they are absorbed or diffused. Thus, an RFZ in the listening room will take care of any floor bounce problem. End of story. Again, it may be ugly, expensive, and inconvenient for listening room occupants in terms of what needs to be done to the listening room to deal with the floor bounce, but it definitely CAN be done with passive acoustical tools, and can be done for the bounce from any number of channels, not just two. Studios physically deflect the audibility of the floor bounce from the front channels by positioning the mixing desk between the listener and the specular (mirror-like) reflection point on the floor. Very thick (say, a foot) floor padding will also work down in the 100 – 200 Hz region of this problem.
Proper Speaker Radiation Patterns
Equalization, even sophisticated digital signal processing, can only efficiently correct the direct sound coming from the loudspeakers, not the sound reflected off room surfaces. This may not be obvious, but it is true. It is one of the flies in the ointment of the efficacy of all attempts to electronically equalize system response using either analog or digital electronics. With most speakers, broadband DSP equalization may help, but leaves some listeners, me included, somewhat dissatisfied. The inability to correct the indirect or room sound is one reason why.
Equalization works best for low frequencies, those below 300 Hz or so. Even with low frequencies, I find the result problematic, with measured flat response in this region usually sounding somewhat anemic. I have to add bass level to restore realism and if I move my head a few inches the bass is not subjectively flat anymore.
Recently, REG of TAS has come to the conclusion that, for maximum realism and reciprocity to the recording, home speakers with multiple drivers should have wide (as in at least wide enough to surround a 12" bass driver) front baffles to avoid baffle-step effects (sudden changes in directionality of radiation) in the midrange. Home loudspeakers should also have constant directivity up to about 4 kHz, above which frequency directivity should increase. REG reasons that only such a radiation pattern can be made subjectively flat using even digital equalization for the lows and common room treatment to absorb high frequency reflections. Only for such speakers can the midrange room reflections be easily minimized in their audible effects via passive acoustical treatments and good speaker and listener placement. REG began his exploration of this topic here, especially beginning at page 16 with the part labeled "The Myth of Wide Dispersion and What Actually Makes Speakers Work in Rooms."
As is typical, REG's conclusions run counter to current audiophile speaker design practice, which either emphasizes narrow front baffles and wide, uniform dispersion to very high frequencies, or which goes to the opposite extreme, producing ever-increasing directivity starting at a very low frequency. REG finds the typical narrow-baffle speakers to be too bright sounding in upper mids and highs even when measuring flat on axis. I agree. He finds those speakers which have decreasing directivity from a very low frequency on up to sound too lifeless.
According to REG, one example of a currently available, reasonably priced (less than $3,000 a pair) and reasonably sized (about two cubic feet—a large "bookshelf" sized) speaker which should be able to work exceptionally well in a home listening room is the JBL Pro LSR 6332. REG has noted that pro-audio folks are well aware of the need for a constant directivity index in the midrange and thus it is not surprising that some speakers aimed at the pro-audio market, like this JBL, are designed to deliver this result. Note in the graphs how the on-axis directivity index is practically constant up to 4 kHz, how well the directivity index of the direct sound follows that of the first reflections, and how closely the frequency response at 30 degrees off axis follows the response on axis, with the off-axis sound falling off in response above 4 kHz.
One might wonder about why other types of speaker designs cannot solve the reproduction problems when used in real rooms. Here are my observations:
- Panel dipoles do not suffer from baffle-step problems, and can produce nice bass detail, but subjectively tend to produce inadequate bass impact and dynamics. I know all about this and it is the major reason I keep coming back to box bass after occasional forays into dipole land. Dipole panels also tend to strongly reflect sound from the wall behind them, defeating even heroic attempts at surface damping unless the speakers are placed many feet from the wall.
- Corner horns produce very strong reflections from the adjacent sidewalls, introducing midrange colorations and bright/gritty sound when measurably flat. The proximity to the wall behind them also severely truncates the subjective depth of field. Heard Klipschorns lately?
- In-wall speakers also suffer from severely truncated depth of field, perhaps because they are right at the wall behind them. It is harder to visualize images behind an opaque wall than when images can seem to happen in free space between the speakers and the wall behind the speakers.
- Full-range omnidirectional speakers can sound exciting and "big," but again produce very strong reflections from all room surfaces, coloring the midrange, brightening the overall sound and fuzzing up image placement in favor of an overall very expansive but amorphous sound field. Some folks like this effect a lot, but, for me, it wears out its welcome quickly.