Jump to content


  • Content count

  • Joined

  • Last visited

  • Days Won


SME last won the day on December 11

SME had the most liked content!

Community Reputation

217 Excellent

About SME

  • Rank
    Super Bass Overlord

Profile Information

  • Gender
    Not Telling

Recent Profile Visitors

414 profile views
  1. Sound card and measurement mic upgrade

    Interesting. Thanks for posting this. I cannot hear any difference on the kick drum sample. On the 50 Hz sawtooth, I am hearing differences, but I suspect this test is flawed. First of all, the most distinct difference is at the inversion points. There is only one inversion point in the sample itself. The other inversion point is introduced by repeating the waveform. I used Audacity to repeat the waveform into a longer sample to avoid any artifacts from the playback program itself, but it had no effect on the sound. Depending on how the waveform was generated, there may be an extra sample at the end that should be omitted when repeating it. I think I can hear a very slight difference in the continuous part, away from the inversion points, but it's really too short for me to tell for certain. The perceptual influence of the inversion points seem to be very strong and have lingering effects. There is another potential problem. The wave form data reaches full-scale in a number of places at both the (+) and (-) ends. A 16-bit digital waveform is inherently asymmetric in the sense that full-scale on the negative side is about 0.0027 dB higher than on the positive side. More significantly, the true waveform peak may actually exceed full-scale leading to an amplification of the asymmetry. Also, it's possible that the waveform was not merely peak-normalized after its generation but was clipped during the generation itself. One reason I have to believe this is that when viewing the waveform in Audacity, the pattern of peaks is irregular about each full-scale peak. This suggests that the inverted waveform is not a perfect mirror image of the original. I could try generating a cleaner waveform, but I don't know how important this is. I would not be surprised if absolute polarity was inaudible in reality, and those occasions where people can hear it are due to unexpected asymmetries in either the source material or in asymmetric distortion in the recording and/or playback system. Edit: The sample rate of polarity.wav 11025 Hz. If the wave is precisely 50 Hz, then the waveform pattern should precisely repeat every 2 cycles because (50 / 2 = 25) divides evenly into 11025.0. This waveform is obviously flawed in a number of ways for the purpose of this test.
  2. Sound card and measurement mic upgrade

    Floyd Toole has a short section on the issue of absolute polarity in Sound Reproduction 3rd ed. in which he cites two studies. One of them appears to be substantially flawed, relying exclusively on LP records. The second, relying on CDs concluded "polarity inversion is not easily heard with normal complex musical program material [... but] is audible in many select and simplified musical settings." However, Dr. Toole also notes that "there appears to be no audio or film industry standards that ensure the deliver of suck absolute polarity sounds from the microphones, through the extensive electronic manipulations in control rooms or dubbing stages, which may be different for different components of a mix, and finally through the playback electronic and loud-speakers at home." Going from my experience, my own playback system has been configured with both absolute polarities, and I've never noticed a different myself, including with bass.
  3. The biggest advantages for multiple subs arise when the different subs have overlapping frequency response. However, the full benefits can only be realized when the different subs have similar output capability. The advantages in this case are: increased output over one sub alone, and smoother in-room response. When the two subs are mismatched in output capability, then you can still achieve smoother in-room response but only by substantially wasting the capability of the stronger subs. That's because the smoother response requires the different subs to be run at approximately the same level, and the weaker sub will reach its limits at much lower level than the strong sub. There are some minor benefits to your usage of different subs for different ranges. For one, the different subs can be optimized to play their respective ranges. For example, the horn is very good above 40 Hz, and you can get away with building deep bass subs using drivers that don't do mid-bass well. You also will gain headroom and see less distortion when reproducing wide-band sub-bass signals. However, as with multiple subs handling an overlapping range of frequencies, the benefits are greatest when the output capability of the subs handling each range is approximately matched. Though how they should be matched depends a bit on the content you play and on room gain effects. For example a lot of music doesn't extend much below 40 Hz, so the horn may be able to handle that very well on its own and reach quite high levels. For home theater, however, a lot more bass extends into the 30s, 20s, or lower. It'll likely take a lot of output to keep up with the horn. Of course, that may not be a big deal if you don't ever push the levels that high. On another note, integrating subs of different types can be very tricky. You will want to do a lot of measurements and experiment a lot with different distance and possibly EQ settings to get the smoothest integrated response.
  4. Yeah, manual is best. IF you are going to EQ the LaScala by hand, that's still more reason to not use Audyssey because it will basically undo much of your work. And yes, there is quite a learning curve for doing it manually, but I'd argue it's well worth it. IMO, the performance of good equipment is held back considerably without good configuration. The speaker stand may be OK for sub measurement, provided it holds the mic well in an upright position. However, the tripod with boom arm is best because anything near the mic influences the measurement, especially with high frequencies. In the crossover region, you have two transducers in different locations producing sound. If the sound from each source arrives at different times, then there will be frequency-dependent phase shift between them. The phase shift will cause them to not combine coherently at all frequencies. You don't need to do a full-range sweep, but your sweep should cover at least 1 to 1.5 octaves above and below the crossover point. It's helpful to label each sweep with the distance and/or crossover point you used and view them all together to compare.
  5. I have experience with Audyssey, and while one of its key features is automatic setup of delays, my experience is that this function performs very poorly. When using Audyssey, I almost always achieved better results by manually adjusting the distances from what Audyssey gave. I also do not like how Audyssey re-shapes frequency response. It aims for a flat in-room target, which leads to a poor tonal balance including thin sound and weak bass. Setting distance for the sub optimally can be a bit tricky, but measurements are your friend. Do you have a good tripod for the measurement mic? I strongly recommend investing in one. Setup the tripod and mic at ear-level at the MLP and then run a series of sweeps, varying the sub delay/distance around 1 ms with each sweep. Look for the delay that gives you the smoothest frequency response around the crossover frequency that you choose. Once you've found the best distance, you can also try to incrementally adjust the crossover for an even smoother response. Note that the optimal delay may also vary depending on the crossover.
  6. Sound card and measurement mic upgrade

    Reversed polarity is pretty common with a variety of electronic devices and is easy to fix in measurements. My Motu 16A reverses polarity. I don't think it's a big deal, and I'm not aware of any empirical evidence that polarity inversion is audible at all, as long the polarity is consistent for all output transducers. The exception would be for crossovers that are designed to be inverted. Though this does highlight a more general problem when using systems with speakers of different types, even if they are from the same product line. The issue is not really timbre matching but rather phase matching. When the crossovers are not identical, there will be phase mismatch over at least some of the range, and this definitely is audible. I have discovered that multiple speaker integration in a system with biquad EQ capability can be improved by using all-pass filters.
  7. I can agree with a lot of this. I enjoyed the movie a lot, despite it being a narrative train-wreck. I really appreciate the story they were trying to tell, but they did a very poor telling of it. I think there was 30-40 minutes (or maybe even more) of footage closer to the beginning that should have hit the editing floor. In its place, they should have devoted a lot more time to developing the characters. Instead, the plot felt very forced with key character developments being implied "after the fact". The scripting was also sub-par with many lines and much of the comic relief being awkward and out-of-character. I'm really surprised now to see critics liking it so much. My opinions on movies usually side with the critics but not this time. Still, it was definitely worth watching, and I will watch it and enjoy it again, but I cannot call this a good movie for the reasons stated above. It is a shame because it could have been so much more. I hope this director does not come back to the franchise As for the sound, I agree that it seemed to have a lot in common with the previous movies. I didn't get the impression that 30 Hz was emphasized, but it's hard to say for sure without a good reference. The subs in the cinema I went to were decent and delivered balanced LFE with a fair bit of chest sensation and even some signs of extension into the 20s. Other than that, the tonal balance of the presentation was definitely mediocre. The speakers themselves seemed to be nice and smooth. I think they were fairly new model JBLs. The surrounds, which were visible, had nicely contoured waveguides matched to what looked like 10" woofers. The sound was quite thin overall, especially in the 250-500 Hz octave, and it seemed to emphasize 1.5-2kHz a bit too much while also sounding too rolled off at the top. That's completely consistent with what I'd expect in an X-curve calibrated cinema, and I reckon the re-recording mixers applied EQ quite sparingly on the dub-stage. Toward the end of the movie where things got a tad louder, the ~1.5-2k emphasis was slightly fatiguing. Oddly enough, the surrounds sounded significantly better, which made me glad that they got used a lot. I think they may have mixed the score a lot more into the surrounds this time than in TFA. The funny thing is that my wife and my sister and her husband could easily tell the inferior tonal balance as well. We all watched TFA on my system last Sunday. I'm real stoked to get this one home and hear it with some BEQ. I'm sure the sound will really come alive then. I have a feeling that it will have better bass than TFA did. I dunno if it will compete with Rogue One + BEQ though. That one was really awesome with the stupid 30 Hz hump fixed.
  8. The Bass EQ for Movies Thread

    I recently tightened down my latest system EQ config, including a complete overhaul of the surrounds that delivers stronger mid-bass and more bass overall. It's nice and punchy for music, without compromising deep bass, where it does show up. I did some testing with music mixed to mono and sent to the center and each surround to confirm that the mid-bass retained its punch on each channel. Over the last few days, I've been testing with movies. The opening bits of GOTG2+BEQ are even better than when I watched it before. The kick drum on the music tracks has life! Tonight I watched "Star Wars: TFA" again with BEQ. I tried with the full mid-bass boost in the BEQ, but backed the PEQ gains down to only +2 dB per channel and added about +0.75 dB @ and below 30 Hz . With the full +4, the mid-bass boost overpowered and killed the deep bass, but it obviously lacked punch at only +0. The extra +0.75 dB down low seemed to get things just right. There is a great balance of shaking effect and lots of chest thump. I can't guarantee these adjustments will do right for everyone else being that they are quite small. In any case, the movie was a fun ride. It was the first time my sister and her husband had heard my system since I got the new speakers and subs. They were smiling pretty big when it was over. Now we're all properly ruined before we go to see "The Last Jedi" at a cinema next weekend.
  9. It looks how I would expect with a Q around 0.7 or so. Every sealed sub essentially rolls off naturally at 12 dB/octave in the low frequency limit, without EQ (and of course, more voltage and power demanded by said EQ). Excursion always increases 12 dB/octave vs. frequency for the same SPL output. So once you go low enough, excursion basically remains the same without EQ. Of course, the roll-off will likely be much less your room. I gather that's a single driver? You were going to use 3 of them? If so, you are hopefully modeling the single driver in 1/3 the total volume. Now that you've done that, try modeling other drivers in a similar volume and similar voltage. (See my earlier posts for how to calculate the "wattage" that WinISD Pro expects from actual voltage, and visa versa.)
  10. The main thing that WinISD Pro and other simulation programs will tell you is the best-case max output output from the sub + amp system, at 1 meter ground plane. It's merely best-case because a driver that's pushed near its excursion limits will start to compress and deliver less output. That's not really easy to model, so it can help if the driver has actually been measured in a cabinet, like Josh Ricci does here. WinISD Pro can also help you compare different drivers and different cabinet sizes. As for how much Xmax is enough, well that ultimately depends on how much output you want and whether you'll give it enough amp power to use that excursion. And unfortunately, the ultimate output you get depends substantially on what your room is doing. There're just too many variables to give you a clear cut answer. If building a ported sub, a lower tune will allow for lower extension, of course, but at the cost of output above the tuning frequency. Lower tunes also generally require larger (and heavier) cabinets. So there are trade-offs. If you know you're going to have more than enough output and have no space or weight constraints, then definitely go lower. Some people build huge cabinets tuned as low as 12 Hz. (A few aim even lower than that.) The B&C21SW152 is a solid driver choice for a ported box. Better yet, lots of people have already built ported systems around it, so you can avoid a lot of design hassle by looking at work that others have done on here and in the AVSForum DIY section.
  11. No, the X curve is not applied to the mix by default or in any kind of automated fashion. Instead, the X-curve imparts a tonal shift that affects what the re-recording mixers hear and influences the EQ they apply on the dub-stage. The mixers are likely to boost the highs and lows to compensate for what they hear. What you really should be saying is: room A acoustics =/= room B acoustics. Size is only one of many room variables, and in fact, listening distance and speaker dispersion pattern are probably at least as important. In some ways, this gets us lost among many details be especially important here. A crucial issue is to distinguish between the effect the above variables have on *perception* from the effect these variables have on the *metric* used in the calibration process. Ideally, the calibration process would rely on a metric that is 100% consistent with perception. Power-averaged response, which is the metric used for X-curve calibration, is not very consistent with perception at all. It is, however, strongly influenced by room acoustics. I'm assuming your response is with regard to the fact that music production doesn't rely on standards? Therein lies a real irony about the cinema standards. It is a case of "no standards" being better than "bad standards". The lack of standards in music forces engineers to adhere to established precedent, which serves as an informal standard. They listen to recordings they consider to be good references and mix and master to achieve approximate parity with those references. Dr. Toole calls this "The Circle of Confusion" for good reason, but in fact, I'd argue that the situation with cinema is worse. That's because, while on the one hand, the cinema standards fail to achieve consistency between different playback systems, the engineers trust in the accuracy of their "calibrated" systems and mostly disregard precedent when making mixing decisions. They simply mix to "what sounds best" to them and assume it will sound like that on other properly calibrated systems. Now to be fair, not all cinema engineers are mixing like I describe above. Through their experience, they have surely noticed that different dub stages sound different and have learned to compensate their mixing technique accordingly with the aim of achieving better results in a wider range of venues. Furthermore, the X-curve standard was actually a decent even if imperfect 1970s-era solution to a very real problem: high frequency absorption of screens is variable, and the best calibration tools that were available at that time relied on power-averaged response analysis of pink noise signals. It's just that today, we have much more capable measurement methods and a much better understanding of perceptual issues. Along those lines, I disagree that Dr. Toole's recommendation (see the second of the two above papers) for calibrating in-room magnitude-smoothed response to a standardized sloped target is the optimal solution, but I believe it'd be a big improvement over the X curve. His recommendation would effectively free up an extra 4-6 dB of headroom per screen channel in cinema soundtracks and would probably lead to a big improvement in the bass for cinema presentations overall. (I'll refrain from giving a detailed justification for this final point unless someone wants me to.) You're right. I didn't have to expand into great deal. I'm just a big geek, you know. And I'm actually quite excited because I think I've finally mostly unraveled a lot of things about film audio that were previously confusing to me and still are confusing to many others. I stand by my statements about the X-curve standard inhibiting headroom on cinema soundtracks, but in time, this is becoming a lot less relevant for those of us who mostly care about home theater, because home mixes are becoming more and more common and are improving in quality. I would not be surprised in the least if "Dunkirk" is a clear exception to this trend. It's probably a straight-up cinema mix and a very loud one at that. Which is still fine by me because I'll re-EQ it as needed when I get a-hold of it. The X-curve is still a big problem in cinema, and I think it's hurting the industry, even if they don't realize it or won't admit it. Dr. Toole has pointed out that many cinemas are hosting music and sports events and corporate video conferences, and stuff like that in order to bring in more revenue, but because they are calibrated on X-curve, all that other audio sounds like poo. That can't be good for their bottom line, and it's the kind of thing that customer satisfaction surveys aren't likely to reveal, being that the influence of audio quality is so unconscious. FWIW, you're like one of the least "asshole" kind of people on these parts, which is why it's kind of funny the way you responded to me. Often that kind of thing pisses me off, but I don't care at this point because I know you and because it doesn't matter that much anyway. Part of my confidence regarding the X-curve is that I can clearly hear it. I'm routinely identifying cinema mixes and re-EQing them to sound better. Ahh yes, so now you think I'm blabbering in Audiophilese? "I can hear the difference man! This will totally transform your audio experience for the better." OK fine, but consider that I really suck at understanding dialog in films. Like, my ears aren't golden at all but are tarnished, maybe even rust colored, right? So when I apply re-EQ and dialog that was shouty and muddy and almost impossible to follow suddenly becomes clear and intelligible, I take note. That's what I'm talking about here. If you'd like, name some titles, hopefully at least one of which is in my library. I'll put it in and try to identify if it's a cinema track that will benefit from re-EQ, I'll play around with it and then publish some PEQs to try to see if it cleans up for you. Is it worth a try? Otherwise, come visit me here in Denver and hear for yourself.
  12. Bulding the Room2 listening room

    I'll concede on this point. Though some domes are quite capable, and if they aren't crossed too low, you have to push things pretty high to reveal the difference at typical domestic listening distances. I agree they would hear differences, but not just in the soundstage. Really, any substantial differences in linear frequency response, on and off axis are likely to be heard.
  13. Here's a recent paper documenting B-chain cinema measurements and attempting to assess the side-effects of X-curve calibration: https://www.smpte.org/sites/default/files/SMPTE TC-25CSS-B CHAIN FREQUENCY AND TEMPORAL RESPONSE ANALYSIS OF THEATRES AND DUBBING STAGES 1 Oct 2014.pdf Dr. Toole was involved in the work above. Here is Dr. Toole's follow-up paper, which suggests a way forward for a more reliable, universal calibration method: http://www.aes.org/e-lib/browse.cfm?elib=17839 Unfortunately, both of these papers are significantly flawed, and I admit I've been slacking as far as communicating with Dr. Toole to explain the flaws. The major issue is that all the "frequency response" measurements rely on magnitude smoothing, which even at 1/48th octave resolution has side-effects that are unexpected to the people doing this work. As noted in my previous post, magnitude-smoothed response measurements are very different from power response or power-smoothed measurements (which the X-curve relies on), and I am of the opinion that neither consistently correlates with tonal balance perception. Hence, the notion that calibration of magnitude-smoothed frequency response in all cinemas and/or all homes to a single target curve (even if it's different for cinema vs. homes) will achieve translation is fundamentally flawed. I have been working on a solution to this problem using complex-smoothing, i.e. FDW, but it is still a work in progress and has considerable limitations. The method requires measurements at multiple locations, even when the goal is to optimize sound at a single seat. That's because reflected sound arrivals do influence perception. To achieve true single-seat optimization, I would have to develop more sophisticated time-frequency methods, and to get it right, I'd probably need to do proper psychoacoustic studies for which I don't have the resources to do right. I don't know. I haven't even taken the first steps down that path. After my various refinements and tweaks, I am getting superb results using my FDW method, but it is tedious and probably error prone for many situations. I still have to make some manual judgments and "fudges" to deal with interference issues involving both my center and surrounds. It'd be seriously awesome if someone would pay me to actually be solving these problems. That'll probably happen ... in my dreams, heheh.
  14. Do double-check with your friend, but I'm quite certain that calibration to the X-curve is still standard for cinema. And that is true despite better screen materials and compression driver tech. Yes, it absolutely does kill headroom, effectively, when compared to how things work for music and "home" mixes.. The 4-6 dB figure is a rough estimate on my part. The cinema basically calibrates for a flat power response, even though a flat speaker under similar circumstances is likely to develop an in-room response with much more tilt. This approach is flawed from the outset because research strongly supports flat direct sound response as preferred over flat power response. While music production does not rely on any standards, the established precedent for music is flat direct sound response, and I am convinced that targeting flat direct sound response leads to better "translation" as well. For treble, a 1 dB/octave tilt is likely to be typical for a flat direct sound speaker (that is, flat after compensating for screen effects), so the X curve is tending to attenuate the top an extra 2 dB/octave above 2 kHz and even more above 8-10 kHz. That's lost headroom because the content in the soundtrack is boosted to compensate. However, there is attenuation at the bottom too, and it is has more severe consequences on headroom. Research suggests that even in fairly dead cinemas, direct-sound-flat speakers exhibit a substantial power response rise, starting below 500 Hz or so and rising more rapidly below 200-300 Hz. This probably occurs in part because of the relatively poor bass absorption of typical cinema room treatments, but perhaps more important is the considerable drop in directivity exhibited by almost all cinema speakers (and consumer speakers too) in that region. Unfortunately, the amount of attenuation during calibration likely varies a lot more than at the top. In fact, evidence suggests that front left/right vs. center vs. surround speakers in the same venue often sound different after calibration. As a consequence, cinema mixes translate poorly even to other cinemas. Cinema sound will never be "great" as long as this broken standard remains in wide use. So because of the attenuation performed during calibration, the low frequencies are boosted on the track also, and unfortunately the amount, center frequenc(ies), and shape of the boost(s) are inconsistent from track to track. Based on the measurement data I've seen as well as what "sounds right" to me when I do my re-EQing, I'd guess that low-shelf boosts of 3-5 dB, centered between 200-300 Hz, are probably common. So that's 3-5 dB lost headroom for bass where the majority of the energy of a soundtrack lives. Additional boosts centered at lower frequencies may also be introduced, but I believe those are less common and less likely to impact the quality of the track anyway. The stuff in the 200-300 Hz range has a big impact on the quality of the dialog, so this adjustment is almost always made, in conjunction with the adjustments made to the highs. Given what I describe, I would argue that 4-6 dB is a totally reasonable figure for lost headroom. When that same track is remixed "for the home", it is remixed on a system that lacks attenuation on the high end and is less likely to be attenuated on the low end either. The system is likely to use either quality flat-direct-sound monitors as-is (with bass response problems being address primarily through acoustic treatment instead) or calibrated to a curve more similar to Harman's recommendations, i.e. with a significant slope. The level calibration is done using pink noise band-limited to 500-2kHz, allowing the rest of the power response to either "fall where it may" or adhere to a non-flat target. (Actually, it's usually magnitude-smoothed response not power response that's fit to a target. Sadly, few people in this business realize that the difference between these two is significant.) Argh, see what you made me do? I wish I could explain it more easily. The gist is that the X-curve standard is flawed and leads to somewhat consistent attenuation of the highs and very inconsistent attenuation of the lows during calibration, and the compensation performed during the re-recording mix costs headroom.
  15. That's not important. With EQ, you can reshape the frequency response to be practically whatever you want. However, this has no effect on the ultimate capabilities of the sub system. If you need (for example) a 30 dB boost at 10 Hz to make it flat there, you most certainly can do it with EQ. However, as soon as you play anything remotely loud at 10 Hz, the amps are going to clip or the sub is going to bottom. The point of the simulation is to see the max capability of the sub, with or without EQ. The EQ feature in the simulation is likely to mislead you as to the sub's capabilities. Sorry, I wrote a typo in my post. I was trying to say that 130 dB SPL down to 30 Hz is not unreasonable. Likewise for 120 dB or higher @ 15-20 Hz. Those numbers may appear to be extreme, but they aren't far off when reproducing many movie soundtracks at a spirited / reference level. For a ported sub that's intended to extend low, high Xmax remains important, but motor force becomes more important than for sealed in a large volume. Investing in pricier drivers with a bigger motor is usually worth it.