Jump to content

Bossobass Dave

Members
  • Content Count

    899
  • Joined

  • Last visited

  • Days Won

    82

Everything posted by Bossobass Dave

  1. Higher res is good. You can e-mail them to me and I'll save them in a folder until I run the waterfall graphs: bossobass@bossobass.com Also, is there any way you can do a slightly heavier trace line? The red trace is hard to see, although that may change with a higher res copy.
  2. Working on spectrograph caps for WOTW, Transformers and Transformers II. I may do more if time permits. For the ratings, I'm changing the Template to Level, Extension, Dynamics and Execution. As soon as Max gets the numbers, I'll load the stars.
  3. Found this sleeper. Assassin's Bullet with Donald Sutherland and Christian Slater. It's a pretty basic foreign intrigue flick but I found it better than a lot of the drama drivel the wife brought home to watch lately and the low end surprised me. So, thought I'd add it to the list while I was at it:
  4. I was wondering if you used the DVD or BR of Thor? I found the DVD to be quite normal and the BR to be like 10dB hotter. Since that was mic'd at the LP I didn't attribute it to the difference in players (I use Oppo for DVD and Panny for BR). I guess most don't do a rent the DVD then buy the BR or if they do they don't notice a diff, but this one was strikingly different, actually potentially system-damaging different, depending on listening habits and system.
  5. Nothing beats relevant data. In the past couple of years I've had a few discussions with the recent proponents of low Le, top heavy drivers. The #1 reason has been making these SL graphs of scenes from every subwoofer-required film in my library. A striking theme in these graphics appears soon after studying them that, after a long while, just jumps out and slaps you in the face... Look at the relative weight differences in peak and average and snapshot between the 1st 4 octaves (3-48 Hz) and the top 1-1/2 octaves (48-120 Hz). This clicked with something I read way back from Dolby in which it was recommended to filter the LFE at 80 Hz because the 120 Hz filter was a brick wall filter, so you would want your effects to be over before they hit that brick wall. Since then, others have said that the LFE channel is not necessarily brick walled at 120 Hz, but you can most certainly see that most soundtracks (of interest here) have a no mans land from 90-120 Hz. This drove me to reconsider the added cost of a flat-to-500 Hz driver motor in terms of $ and compromises to the actual goal. Max's graphs drive it home in a blink. I LOVE it when some really smart guy aims his guns at a problem and brings the results to the rest of us to see things we otherwise would never see. Awesome stuff, Max and thanks a truckload. I can't tell you how aggravating it was for me to suffer the last go-round with the "the graphs aren't scientific, the content is garbage, I've heard it and it's irrelevant", and my all-time favorite, "Bosso is trying to sell subwoofers so he HAS to say it's important" ... bulllllll shit. It's a ton of work to dig up actual facts regarding the physics, industry members' takes, Dolby and Holman's input, designing and building a system to reproduce faithfully, setting up software like SpecLab, running and posting hundreds of graphs along with the facts, etc., etc. What for? Certainly not for me because I already know this shit fairly comprehensively. Then to get pissed on for keeping the general consensus on track by folks who've never posted a single shred of data for anyone's benefit. Anyway, as Max said and as is usually the case, from the bad springs a better version with no noise frem the peanut gallery. Anyone who wishes to ignaore the facts can simply head back to the junk yard where the rest of that sort hangs out. I agree that it was so appropriate to land it here where Josh & Kyle have done a bang-up job providing data. This sort of exercise ties in perfectly with what they've provided and hopefully it will be an ever-evolving exercise to the benefit of everyone. BTW, I agree with Max, log scale gives too much weight to the first decade. I've tried it several times and just find it less usable than the linear scale. I'll post an example comparison when I get a second to be sure everyone else agrees and it isn't just my own bias.
  6. Got it, and bingo: One thing: Change the sampling rate back to 48k Hz. 96k Hz with a decimation of 4 (my setting) gives a sampling rate of 24,000 Hz. What that does is spread the resolution across twice the number of frequencies, which is why the graph is pixelated, like my old settings were (which is when I graphed the scene in the above pic for comparison). If you change back to 48k Hz, it should match mine perfectly. The only other thing is that with a 0-480 Hz scale, there are no division lines to tell you what frequency the content is. It would be perfect if there were an option for the grid in the waterfall to be in 30 Hz increments, but there is no such option, so for now, you have it as good as it can be. It's just a glitch thing (the frequency scale) that I don't know how to deal with yet. If you could change the sample rate to 48000 and run that scene 1 more thime, I'll appreciate it.
  7. This is too cool. Just too cool. As for the <7 Hz, I can always run a direct digits spectrograph and extrapolate that part of it. It's only 1/2 of the 0-10 Hz leg, and I would only have to do it if there is actually any content there to look at in the first place. I'm going to go direct-digits anyway to eliminate the "you use a mic" whiners. That means I have to redo my entire library because the ones I've graphed (like Star Trek, Hulk, WOTW, X-Men 1st Class, etc.), will have to be in our list. Either way, this is shaping up to become one heck of a standard for reference.
  8. I agree that we should keep it bass-centric. I have lots of movies (including The Natural) that I love to watch for various reasons, but they aren't ever gonna be graphed for this thread. If we keep it bass-centric, then rent/buy/no buy is great. For example, I loved Prometheus, so I bought it even though the low end is anemic as heck. (Yeah, i saw others post at AVS that it had amazing/powerful bass, etc., but, if Max graphs it, it will become evident that that ain't so). That means there are indeed other factors that would go into that rating. I vote we put you and Ethan in charge of this, after seeing the awesome job you did on your shootout thread.
  9. Thanks for the effort, Larry. I'm bringing this one here (have already e-mailed you on this) because it's a head scratcher for sure: I put my digital copy of this scene up in Omnigraffle with your latest posted version. I tried to reconsile the 2 graphs for a long while and could not do it. For example, you show a lot of content in the 60 Hz area that just isn't on the disc. Finally, my brain caught it. I stretched my graph until just my 0-30 Hz lined up with your 0-120 Hz frequency scale and there it was, a perfect match (although my graph is distorted to a large degree): Your graph is on top, my version is on the bottom and the severely stretched version is in the middle. I have no clue as to how your SL frequency scale is so far off. I suggested in e-mail that you change your frequency scale (top left corner of the display) to (I think) 0-450 Hz. I would actually like to see 0-480 Hz (reasoning that if your 0-120 Hz is actually 0-30 Hz, then if you quadruple your scale to 480 Hz, you may actually capture 0-120 Hz). Windows (ugh) I also suggested you reboot your computer and try the graph again. If you can post the expanded FR scaled graph here, that would be great...
  10. I understand, but I was just saying that the I/O decimation still has to be entered manually or there's nothing to account for. No, is perfect. :^)
  11. Remember that decimation works on 2 fronts, FFT and sample rate (found in Audio I/O). FFT resolution gets washed out when it's spread across 24,000 Hz (48k Hz SR, decimate=1). Freq BW will = 1/2 your sample rate. Just making sure you know that. Also, we need to get a rating system down or I have to stop posting the SLs because I don't want to have to go back and change all of them after the fact, so I'll put this on hold until then.
  12. Just updated by adding Prometheus below CITW above. What a snooze LFE-wise. I guess if you run the sub +10dB hot it might be good, but I haven't tried that yet.
  13. I like the 2nd metric, by octave. How are you getting the feed into SpecLab? I'm assuming you're using an analog connection from the AVR to the sound card, thus the roll off. Any way you can connect your player to the computer via USB? Or, if you have to suffer an analog stage roll off, if you have analog outs on your player you can connect that way and avoid the AVRs roll off, which I'm assuming has to be the offending roller. My interface has digital coax in and my player has digital coax out. When I go direct, I bypass all analog stages and get the 1s and 0s all the way down. Comparing WOTW plane crash, digital feed vs mic'd version:
  14. There are as many windowing choices as there are application for them. Unfortunately, they aren't all available in the SL drop down menu (such as Blackman'Harris, Blackman/Nuttall, etc.). Of the choices available, I chose Nuttall for it's increased dynamic range over Gauss. Here's a roughly scaled comparison (again, pictures beat words):
  15. Since I mic and go direct, there is no universal setting for calibration. Spectrumlab has a calibration methodology: To realize absolute voltage readings, level readings in dBuV, etc, the program needs to know the relation between input voltage and A/D converter value. This is a rather complex calibration, IMO. I see no reason to go through that process for waterfall graphs. The problem I have is that when I switch back and forth from digital feed from the player to the digital feed from the interface I have to re-calibrate, so instead I use the offset feature to reconcile the 2 different inputs. I set the level from the interface with the right side amplitude bar. Headroom is dictated by the input from the interface. When I switch to the direct feed from the player, I adjust the "offset" to match the interface input by eye, which is accurate to within a couple of dB since I'm very familiar with the color scale and I'm comparing the exact same scene. It's by no means absolutely accurate, but neither is anyone else's method, which is also done by eye, both methods being roughly the same accuracy, which is good enough, IMO. The color intensity setting is also set by adjusting the "offset" option. I take a maximum peak from a scene like HTTYD Red Dragon crash scene and use it as the metric for maximum where any single frequency is approximately -5dBFS. Again, this is by no means absolutely accurate, but is based on viewing thousands of scenes compared to max digital RS meter readings, system reaction (VPL, clip limiter indicators vs output voltage readings) and all averaged in general. It gives me +5dB of "headroom" because I set my color scale range from -60dBFS to +5dBFS. I don't believe it has ever been shown that there exists a scene in any soundtrack the scenario where 7 channels plus the .1 channel all simultaneously contain coherent 3-120 Hz content at maximum level, so I disregard the 125dB estimate. Of late, the only 2 scenes I would imagine might fit that scenario are the one pictured in the CITW I posted here and the Thor BR scene at the end of the battle with the Ice Giants when Odin shows up. In the latter, I saw a +10dB increase in level vs the DVD. I do not know what that's about. I did use 2 different players to compare that scene, but my system calibration should hold equal for both players, so I'm at a loss as to how the levels are different. IOW, can the mastering house raise the overall level after the disc is encoded, or how else would there be a difference? Some people have reported that my settings, which are in the SpecLab preferred .INI format, do not appear to be the same as my settings. I use Windows 7, so I don't understand the problem, but am currently investigating it and should have suggestions to remedy any problems within a week. Folks who download my settings should use a universal scene and post the resultant capture here, so that I'll know what any problem might be at a glance. This saves a go-zillion e-mails with graphs that need explanation before I might even begin to see a problem. Since most people have WOTW, the lightning strikes scene is the easiest for me to recognize immediately. Begin where Cruise and Fanning are indoors, Cruise is at the sliding glass doors and says, "Don't worry, lightning never strikes the same place twice" and end where they're under the table and Fanning says, "Why doesn't it stop?". Awesome suggestion, which is what I needed. I have far less time than I need to do all of this graphics and measuring stuff, so I'm really looking forward to guys like yourself to jump in and pick up the slack. I agree and it has always annoyed me when guys set the scale and time stamp info in the waterfall capture. Here's the change: As far as a rating system goes, I personally use a simple "BUY" or "DON'T BUY". It includes story, cast, script, visual effects and soundtrack. IOW, movies like Cloverfield do not cut muster. Sometimes, the sound will be good enough to outrank the other categories, but Cloverfield, for all its hype, is weak compared to the top soundtrack discs, so the other categories make it a "DON'T BUY". OTOH, WOTW is definitely a strong enough sound track to tip the scales. So, whatever everyone decides as a rating system is fine with me. I really dislike subjective descriptions and rating systems. It's like describing the color red to a blind person and always leads to a dog-chasing-its-tail debate. Tom Noussaine has that largest single body of eval results I'm aware of. He did all of it in-room, with the flaws that entailed (like his rooms +4dB peak at 32 Hz), but useful as a comparative tool in the end. Keith Yates is the only guy to ever use SpecLab and movie soundtrack scenes to compare subs, but he did so outdoors, GP with no sats (although he did mention a difference in overall dBSPL when FL/FR were added). I've always wished we could get a combination. Do the SL graphs of a global set of scenes indoors. Despite what the conventional wisdom says about all rooms being too different, I've found that to be largely a myth. It's not that hard to get a good enough (+/-) 3dB) FR in-room and if every sub is placed the same and the mic as well, the results would be hugely valuable. Josh, Why can't I drag an image code file onto this forum (the Mac unclunk method) vs copy/paste (the Windows clunk method), and what the heck is the default font/font size? When I copy/paste, the font changes to the pasted font and I can't seem to find the default font on your font options drop down.
  16. 2.25-Star Films: 47 Ronin Closed Circuit Mission Impossible II Windtalkers 2-Star Films: Afflicted Mission Impossible Toy Story of Terror 1.75-Star Films: Only Lovers Left Alive SuckerPunch Edit by nube: First post got too long, so I had to move maxmercy's measurements methodology here. Measurement Methodology: The Overall Star Rating is an average of the following four categories (each one explained in greater detail below): 1. Level 2. Extension 3. Dynamics 4. Execution 1. Level - This is measured by digitally bass-managing the 5.1 or 7.1 signal. The Level is a composite number, and is calculated by the average of the [1] highest peak in dB (maximum 126dB for 5.1, 128dB for 7.1), [2] the average/RMS dB level of the track (125ms integration time) and [3] the RMS peak level (loudest single 1/8th of a second of the film) in dB. Full modulation of the waveform is considered to be 0dBFS. The ratings are as follows: 5 Stars - >112.5dB composite 4 Stars - >110dB composite 3 Stars - >107.5dB composite 2 Stars - >105dB composite 1 Star - <105dB composite 2. Extension - The same bass-managed digital signal above is analyzed with Spectrum Lab, with Nuttall windowing, and either a 1 second or 2 second integration time (the longer the integration time, the higher the resolution, at the cost of dynamic peaks, but for the extension category, it is a moot point). Extension is determined as the -10dB point from the overall peak of EITHER the Peak (green) or Average (red) trace, whichever is higher. The ratings are as follows: 5 Stars - <10Hz extension 4 Stars - <15Hz extension 3 Stars - <20Hz extension 2 Stars - <25Hz extension 1 Star - >25Hz extension 3. Dynamics - The Overall RMS/Average Level of the track from the Level category is subtracted from the Overall Peak Level of the track from the Level category. The ratings are as follows: 5 Stars - >27.5dB Dynamics 4 Stars - >25dB Dynamics 3 Stars - >22.5dB Dynamics 2 Stars - >20dB Dynamics 1 Star - <20dB Dynamics 4. Execution - This is the only purely Subjective category, and will be decided by a plurality in a poll, with single star increments available to be voted on. Vote what you thought about the LF use in the track, and note any clipping/compression/filtering that took place Please note that no objective measurement system is perfect. The way the measurement scheme is set up, it will be very difficult to achieve 5 Stars overall, and very few films have done so. How good a movie is to watch has no bearing on how it measures. This is a way to objectively look at the low frequency content of a track without all the subjective "it was awesome" and "seismic" rhetoric from professional reviewers for nearly every BD release. There will be some films which are LF monsters that will not garner 5 Stars overall, and everyone has their favorites. In toto, you can expect a film with a higher Overall Star Rating to have more LF than one with a lower star rating, and most of the 4+ Star films/tracks are very good LF experiences. **** Beginning of Bossobass's post here **** Using Examples (not necessarily with correct time stamp, ratings, pics) of movies I've recently seen, here is where I'm at so far. This will be the template and all graphs and pics will be sized to fit the template. The top template will appear once at the beginning of the title and the bottom template will appear for each successive screen cap with the proper begin/end time stamp and begin/end pics for each scene. First, for those who have asked me how to interpret the SpecLab waterfall graphs, here's a quick how-to: The waterfall scrolls from top to bottom as the scene plays, so the beginning of the captured scene is at the bottom of the waterfall, which is documented by the time stamp and a picture of the scenes beginning and, of course, the end of the scene is at the top of the graph, marked by the time stamp and a picture of the end of the scene. Per Josh's suggestion, there will be 3 scenes for each movie. I may, from time to time, post additional scenes if a movie is exceptional or because someone has requested to see the content of a scene, but I'll post them separately to keep the archives of the films uniform. If anyone would like to capture scenes in SL to help out with the workload, I've put Spectrumlab and my settings, as well as instructions for loading my settings into SL and the above pictured 'how-to-read-a-SL-waterfall-graph graphic on my site: http://www.bossobass.com/Bossobass.com/Technical%20%28cont%29.html Make sure you note the beginning/end time stamp numbers and take a pic off your screen of the beginning and end. So, for each scene you need 2 pics, 2 time stamps and the waterfall graph. The movie pics are easy to get online at IMDB, which will also have the sound designer and mixer info. Members who create SpecLab Caps can forward the caps to me and I'll fit them to the template and return them via e-mail in a file or host/post them. Decide on how the rating will work, which graph (just the waterfall or the 3 graph cap), etc. from here. Any Mac users, I'll be happy to forward the template for Omnigraffle because having someone to pick up the slack is always a good thing. Omnigraffle makes it very simple to size, place and export the template with everything needed to keep the graphs all the same size and resolution. I would also prefer to keep the settings the same for all caps using my settings. They're found in a downloadable file on my site with SpecLab. Just download and then load the settings.INI file into SpecLab. You'll have to make sure your volume settings and sound card settings are compatible/correct as well as setting the offset to match the input of your player or AVR to get the intensity of the colors correct and to insure against clipping. This beats the snot outta fighting the extremely steep SL learning curve. Of course, I prefer my settings. This is because they are the result of years of experimentation to optimize them for the subwoofer range in a scroll speed that is meant to capture scenes, not huge time blocks. This makes a very accurate combination of resolution at the extreme low end in scenes (which is what everyone always asks for a graph of) that can be used to compare with other soundtracks at a glance. Imagine if Josh formatted every set of data for his subwoofer tests differently, using different smoothing, distances, graph scales, etc. It would be infinitely more difficult to interpret the data. Because the data is consistent and presented in a consistent format, it's very easy to use for comparison within the body of work. Anyway, this is a lot of spare time out the window. If done correctly and used in conjunction with Maxmercy's P2A (Peak To Average) graphs, also formatted consistently and in the future in a format that allows for instant comparison with other movies, this would become an extremely useful databass, to coin a term.
  17. I think we're maybe talking about 2 different phenomena. In any case, stored energy doesn't stay in the box, otherwise the box would eventually explode. The length of the sound wave is irrelevant.
  18. IMO, inductance roll off is irrelevant in that you can shape the signal to change that. Inductance is relevant in that it causes harmonic distortion where it's audible and changes the sonic sig. Of course, the number one problem in all of home subwooferdom is playback level. There is far too much emphasis put on how "hard" the sub "hits", "punches", "slams", "pounds", etc. Seriously, WTF does any of that mean anyway? It means how loud can you push the hardware before it pushes back. Sort of like asking asking a guy how fast his car can take a corner as they're towing it out of the canyon. If the system is properly sized and powered for the job, harmonic distortion is all but irrelevant and if the signal shaping flexibility is designed to accommodate the desired X-over point, inductance is all but irrelevant. I believe there is one area that hasn't been properly explored regarding a measurement methodology and a metric. Latent release of stored energy through the surround/cone/dust cap. Years ago I asked gurus like Wiggins what effect the cone/surround/dust cap design/material had on sonic signature and the immediate and unanimous reply was "none". Rubber surround vs foam vs accordion. Coated non-pressed paper cone vs aluminum. Large, stiff dust cap vs no dust cap (a la LMS U). I'm not talking about resonance or break up, just transmission of latent stored energy. This phenomenon affects the sound, one way or another, in every alignment. Is it audible? Does it affect sonic signature? Is it measurable? Is it different from sealed to PR to horn to ported to BP? I have always realized that this is a grope in the dark theory, but so has the entire evolution of subwoofer assessment process been, IMO. Siegfried first turned me on to the LRSE (Latent Release of Stored Energy) phenomenon. It is the single reason he moved on from sealed to dipole. It's the only reason I always stuff the box, because any dissipation of stored energy is a good thing. Siegfried even discussed sintered metal filter products to absorb the stored energy, putting the idea out there and leaving it for others to pursue. There is no question in my mind that frequency response (what the mic tells you you're hearing primarily) is the dominant factor in subjective results and it amazes me that anyone would think otherwise. But, having taken part in many, many discussions of intermod distortion, harmonic distortion, group delay, compression, frequency response and roll off order, etc., I tend to dismiss all of those phenomena based on actual listening and the factual data that's available. LRSE would also, of course, include the enclosure design and construction. I've found that to be quite easy to measure, but have no idea how to quantify it. The drivers construction, OTOH, is a moving target and much more difficult to measure... at least I think it is at this point. Just rambling about a subject that has been banging around in my head like a ping pong ball for a decade. That usually means that there's something there. Just thought I'd throw it out there because the recent inductance jargon has left me pretty unfulfilled.
  19. After reading a bit of yet another "which is better, sealed or ported?" thread at AVS I decided to suggest the onset of a database of component frequency responses. The point I didn't want to waste posting in the AVS thread is that there never has as of yet been a true 'sealed vs ported' listening session, so the question just cannot be answered by citing commercial subs and the slew of cliches that are always brought up in these debates. Several ID companies have offered the statement of performance that goes something like "Our sealed subs have a true 2nd order roll off" Implying "no HPF". This is, of course, a patently false statement without the qualifier of stating a bottom frequency. All EQ circuits and amplifiers have blocking caps to prevent DC offset. The values of those caps determine the F3 and order of roll off, which would be additive to the sealed subs 2nd order roll off. When you include the typical DVD/BR player and AVRs analog SW output roll offs as well as any external PEQ post smoothing hardware, you begin to quickly realize that the sealed sub that is supposedly an unadulterated 2nd order sub is really down -30dB at 5 Hz (or 8 Hz or 10 Hz or 16 Hz or 20 Hz) with a virtual brick wall roll off below 10 Hz. To illustrate, the Epik Empire and the JL Audio are sealed subs and the PB Ultra-13 in 10 Hz tune is a ported sub. FR and roll off are everything and, in this case, ported or sealed is irrelevant. The typical AVR is a Marantz, Denon, Yammy or Onk and there is also a fairly short list of typical players. The miniDSP, Berry, SMS-1, etc., also represents a fairly short list of EQ devices. If we had those loopback graphs, it would be simple to accurately calculate any combination of those FRs to predict the signal chain roll off. The amplifier is a bit more difficult, but not impossible to nail down as well. There is no doubt in my mind that the whole "most guessed wrong at sealed or ported at the blind listening G2G" is the result of much more similar FR and GD than a WinISD model predicts. For some devices (like my external sound card/interface) you want to measure the digital output vs the analog output because they are almost invariably different and the sound card is connected digitally, not by analog out. So, the typical loopback measurement and subsequent correction file may be skewing the results. To measure a players FR we burned a perfectly flat sine sweep to disc. Many players are connected via HDMI and/or digital coax (and no longer even have an analog out), so the sweep should be taken from a reference AVRs SW out and the AVRs roll off subtracted. I still enjoy SACD multi-channel music, so I measured both the digital out of the BR and the analog out of the OPPO SACD player. Since I'm a firm believer in my in-room tests of comparing the direct feed SpecLab of a scene to the mic'd at the LP version of the same scene, I wanted to explore the roll off differences in the signal chain in both cases, since the graphs go to DC and they're directly compared for evaluation of in-room sub performance. Player===>AVR===>Sound Card (no correction file)===>SpecLab vs Player===>AVR===>EQ===>Amplifier/Sub===>Mic (correction file)===>Sound card (correction file)===>SpecLab. The results were very interesting and it led me to think that a database of hardware roll offs would be a pretty important part of evaluations of many a posted so-called in-room FR and "this sub is [ ] than that sub because blah, blah, sealed vs ported". Just a thought. I've done every combination of my own system (otherwise, I would not have had the impetus to extend the response as I have), so it's no big to me, but it's irksome this far into HT technology to hear people praise or condemn an alignment based on a FR that in no way resembles the alignment they're praising or condemning. Having the signal chain chart to quickly predict your systems roll off would be awesome. I'm sure we could put together a simple calc program that showed the resultant anechoic FR by simply clicking on the components in your rack and your sub.
×
×
  • Create New...