|
Transcript Moderator: Welcome to our chat with Sam Berkow. Sam, for those who may not know who you are, could you give us some background? Sam Berkow: I am your typical audio/acoustics geek. I started out wanting to be a musician. I’ve been working for the past 12 years as an acoustical consultant, designing rooms where sound is critical - concert halls, recording studios, etc. I’ve been very active in the field of acoustic measurement as well and I founded SIA Software Company www.siasoft.com, which I sold to EAW two years ago and left last December. I currently work with John Storyk at Walters Storyk Design Group (WSDG) www.wsdg.com. Russel L. O'Toole: What new iterations or new capabilities do you foresee with Smaart? Sam: I am NOT involved with what is happening at SIA currently. Certainly I can see where the DSP and computer capabilities are headed. I think that in the near future, the measurement system should be more aware and interactive with the sound system. Jamie from Toronto: What is the most common mistake you see when people design rooms? Sam: I think the most common mistake in both large and small rooms is the lack of LOW FREQUENCY control. In small rooms this is a HUGE problem, particularly studios and home theaters. Designers both new and experienced often forget how hard it is to really control LF energy. The second biggest mistake is forgetting that great rooms MUST be QUIET.. Jamie from Toronto: Is that a speaker issue, room layout, or processing issue? Sam: This is primarily a room design and treatment issue. The miss-use of subwoofers is another of the most common mistakes made in sound system setup-operation. People forget that the gain subwoofers (relative to the LF devices) are directly related to the crossover frequency. It is really important to remember that when you increase the gain in a sub, you MUST adjust (lower) the crossover frequency. Does that make sense to you? Dan: When you say quiet, do you mean ambient noise levels? Sam: Yes. I find that MANY rooms are VERY noisy, particularly at LF. I prefer the PNC (Preferred Noise Criteria) curves, which limit LF noise more than the standard NC curves. When I worked at Artec Consultants http://www.artec-usa.com/ , Nick Edwards and Bob Essert developed the AR-1 criteria curve as a design target for concert halls (it’s near the threshold of human hearing). For me, great rooms should be very quiet. Sam (continued): Room design involves three main parts: Isolation (making it quiet), Acoustical Response (making it sound good) and Operational Support (making it work for its intended purposes). Russel L. O'Toole: One thing I have learned the Hard Way is to not put a lot of low frequency energy in those huge reverberant churches if I want to maintain good speech intelligibility. Sam: Speech intelligibility is a very big topic. I don't think it is a secret that I do NOT like many of the standard "metrics" of intelligibility. STI and %Alcons may be interesting to some, but I think that it is much more important to find parameters that help you. understand what is wrong and how to fix it! I like two such parameters: The direct to reverberant level difference (at various frequencies) and the ratio of decay rates at mid-high and mid-low frequencies. {STI is the Speech Transmission Index. %ALCON is the percentage of articulation loss of consonants}. Sam (continued): In general Russ' comment can be seen as an example of the problems I mentioned earlier. In most large reverberant churches, there is very little acoustical control at LF, so the decay is tonally unbalanced (the LF often decays at less than 1/2 the rate of mid-high frequencies). Also, if the subs and the LF devices in the system are overlapping for an octave or so, there is a great chance of uneven LF energy throughout the room and MUDDY sound. P. Tucci: Seems that one of the new developments in Smaart is the option of vector averaging in the Transfer Function measurement. Could you do some explaining of that concept? What are its advantages? Sam: I have seen the beta, and I can tell you that I have mixed feelings about it. Vector averaging is a simple concept that has complex implications. Vector averaging is used in SIM http://www.meyersound.com/products/software/sim/sim.htm (which it seems is where the SIA team is trying to bring Smaart). As you know, the transfer function compares the input and output of a system (typically a sound system in a room). Sam (continued): This comparison is done in the frequency domain. In this domain, the DIFFERENCE between the energy in the input signal and the output signal is represented by a complex number, which is in fact a vector. Every vector has a magnitude and direction. A vector describing the motion of a car has a speed and a direction. These vectors have a magnitude (amount of energy) and a delay time (which we represent as a phase angle). When you average a series of measurements, you can use several mathematical techniques to average these vector values. Sam (continued): One way to do this calculation is to just average the magnitudes. This is pretty standard. Another way is to consider the magnitude AND the angle. Using the car example, if you average a car going 60 mph EAST with a car going 65 WEST, the standard magnitude-only average would be 62.5 mph, while the VECTOR average would be "5 mph WEST". Sam (continued): Is this important for acoustic or sound system optimization? I think you can make an argument for vector averaging at LF in very reverberant spaces. John Meyer has said he thinks it’s important, but I have not seen any real value in the field for it. However, John is a smart guy (no pun intended), and I am sure it has some use. Moderator: Sam, you think it's more an academic exercise than a "real world" tool? Sam: I'm not sure. For me, it has little value, but others seem to want it. In terms of Smaart, I think there is clearly a desire to make Smaart more like SIM, which I do not understand. Jamie from Toronto: I have spoken with people on EQ'ing. Some say use it, some say as a last resort… What is your stand and why? Sam: This is one of the silliest statements people make! EQUALIZERS are tools. You can use them well, or use them badly. Crossovers today are extremely complex, very often with MANY filters. I believe that the goal of proper equalization is to OPTIMIZE THE INTERACTION OF A SOUND SYSTEM WITH ITSELF AND IT’S ACOUSTICAL ENVIORNMENT! Sam (continued): I believe that EQ is very often miss-used. I rarely use graphic EQs, as parametrics provide much more control and much better filters. I also think that it is VERY VERY VERY hard to equalize a sound system properly without a good measurement tool. The role of the measurement system is to help you understand what your ear is hearing and how WIDE the resonances in the system are. The Old Soundman: Do you use Meyer http://www.meyersound.com/index.htm CP-10 or other parametrics? Sam: I use a number of parametrics. I like the Apogee CRQ-12 http://www.apogee-sound.com/products/ as my favorite analog parametric. I also like a number of the digital system parametric equalizers. I own a CP-10 and it sounds good, but it is very touchy to use and does not have some features I like. You may be surprised to know that the filters in MediaMatrix http://aa.peavey.com/ seem to exactly mimic the CP-10 filters. Dennis Bohn of Rane http://www.rane.com/ wrote a GREAT AES www.aes.org paper on EQ design about 20 years ago. Bruce N: What are the first things you notice when you go into a new room, before any measurements are made? Sam: I ALWAYS listen first. If there is a sound system, I listen to as little of it as possible, perhaps one speaker or monitor. If there is no system, I clap and yell. (Yes, I look geeky doing it!) I pick music that is very open and sparse... I have a sample of a Shawn Colvin tune that starts with just a kick and a bell. I listen for the relative decay of the LF and mid-highs. I also listen for reflections and noise. Jack Wrightson: We all know that digital correction (measuring the speaker and then implementing the "appropriate " filters, time correction, etc.) can improve both the measured and perceived performance of a loudspeaker system. Does this technology transfer over to larger arrays of loudspeakers, i.e., can it be used in these applications, if so what are the problems, if not, why not? Sam: Currently, I use Smaart to help me understand what I am hearing. I look at transfer functions on axis of speakers and clusters, one at a time and then in groups, to see the individual responses and the effects of groups of clusters. Jack designs many arena systems with 6 or more arrays. If you just EQ to one array, you will be in real trouble as the off axis response of the other arrays plays a huge role in what you hear. Sam (continued): There are currently some "digital correction systems" that will help correct or EQ some systems, but they seem to work best in controlled acoustical environments, or at least in rooms that are not very reverberant. For such systems to work for complex multi-cluster systems is possible, but it will require that the user enter a ton of data about the sound system components, and perhaps even about the relative geometry of the measurement mic(s) and clusters. Sam (continued): Further, I believe that to be effective the system will have to be aware of the speaker types and characteristics, and all processor settings. It seems to me that one could make measurements and choose what to EQ by listening and looking in less time. This is the rub. I believe that such automated EQ systems will require a set of target curves and tolerances (with the tolerances in both magnitude and phase). Sam (continued): In really large rooms, like the arenas and stadiums that Jack often works in, there is another problem. And that is that the transfer functions that we use (in SIM and Smaart) provide a LF time window of about 3/4 of a second. This means that in many cases there will be LOTS of energy outside the LF time window, making it harder for an automated system to select proper EQ filters. Brendan: Can you tell us what/any new services or products you are working on? Sam: As always, I am designing rooms and optimizing sound systems, and I’m currently working on a number of projects including a new concert hall/recording facility for Jazz at Lincoln Center; a synagogue with a major organ and electronic enhancement system (in NYC); and a number of other rooms. Sam (continued): In terms of products, I am currently looking at several ideas, both software and hardware, and will announce things in the not too distant future... I am also doing some writing, and recently cracked a rib skiing! Chris Kathman: Watch those vectors! Sam: Skiing is a definitely "vector" sport! Annacoick: Sam, what's the biggest mistake people make when tuning a sound system? Sam: Good Question. NOT LISTENING! Or NOT THINKING! Or BOTH! I think it is amazing how badly most sound systems are tuned. Gain structure is one terrible issue. The subwoofer issue has become HUGE. I often run subs at + 8 or +10 (relative to LF). I see this all the time. The sub and the LF overlapping for an octave or more, resulting in MUD! All because the sub crossover was set for the sub at equal level to the LF, and raising the sub gain has increased the frequency of the EFFECTIVE crossover point. What a mess! Sam (continued): Also, I hear delay systems that are WAY out of time. I find this TERRIBLE! Lastly, I often see people trying to EQ acoustical or electronic problems, so the misuse of EQs is a big headache. One system I saw had a mid driver out of polarity, so the operator cranked the EQ up to +12 to fill the hole! I also see people building crazy speakers arrays and aiming speakers in VERY strange directions. Tucci: Any preference that you have, Sam? A speaker phase-aligned through crossover or an impulse alignment? And of course, the follow-up why? Sam: This is one of the questions the current pro loudspeaker designers are dealing with. The issue is when you have two devices (typically in one box), how do you align two drivers (say a mid and high or mid and low). With Smaart or almost any measurement tool there are a couple of choices. You can align the time domain impulse or try and match the phase. Sam (continued): With the phase you can try and match the exact phase or just the slope. This gets really complicated with horn-loaded devices, which have a "virtual acoustic apex". So the answer is (drum roll please): Whatever works! (OK, so I wimped out.) Certainly the high-end studio monitors all seem to have very carefully matched phase alignments, however they rarely have horns to deal with! Sam (continued): In my experience the key issue comes down to specific devices and applications. By the way, I learned most of this type of stuff from Don Pearson, co-founder of Ultra-Sound, who is amazing with setting up critical crossovers. (For more from Don, go to http://www.siasoft.com/support_drdon.html - Editor) Chris Kathman: Old school! ken m: What other types of music do you listen to when setting up a room? Sam: As I mentioned earlier, I pick CDs with sparse mixes and lots of separation. I have a stack of CDs that I have learned to HATE musically but have heard on hundreds of systems. I try and pick music that does NOT have long reverb trails, and does NOT have highly layered sound. Sam (continued): Here are a few examples: Shawn Colvin – “Steady On” – (great kick bell intro, female vocal, acoustic guitar); Los Lobos – “Kiko” – (nasty bass and guitar sounds and I like the band); Robbie Robertson – “Storyville” – (the only non-layered Daniel Lanois production); Sex Mob – “Solid Sender” - {OK, I don't use this as a test CD, but I like the band, and as the saying goes: SEX MOB RULES!) http://www.sexmob.com/ Moderator: It's time for another question now :-) BTW......SEX MOB RULES! joel: During show time, the microphone reads both sides of the PA at the same time. How do we work around that if we want to change EQ settings? Sam: This really depends on the mic position, and circumstance. When I start to tune a system, lets use a 3,000-seat room as an example, with a left-right system. I start with my mic, on-axis and in the direct field of one stack, lets say house right. I listen, I measure, I might look at each band (Sub, LF, Mid, High) separately. I set my EQ for this stack with the mic in the direct field, on axis of the stack. Sam (continued): Next, if the house left stack is the same, I copy over the EQ settings and LISTEN.... if the two stacks sound the same (with each on separately). I then go on, if not, time to troubleshoot. Assuming they sound the same, I then measure with the mic STILL in the direct field of the house right stack, with BOTH stacks on. This will help me see any interaction between the two. I SAVE THE FINAL ON-AXIS RESPONSE CURVE. Sam (continued): Now we move the mic back to the mix position. If the system is MONO, and the mic not EXACTLY equidistant to the two stacks, the second arrival is seen as a reflection. By making a measurement, you can now compare what you hear/measure at the mix position and what you heard/measured on axis (stored as a trace). Once you see what you have, you can decide what EQ changes you want to make. One more point: In many empty theaters, it is often a good idea to pull out 120Hz to 180Hz during sound check and bring it back when the audience is in the house, as this is the range that people absorb more efficiently than theater chairs. Chris Kathman: Have you used the Ashly Protea http://www.ashly.com/? Sam: Yes. I think the Ashly is a very good “bang for the buck” device. I found the front panel a bit confusing, but I use the Smaart interface to control it. It is not the quietest unit around, but once again, certainly very good value. joel: During a live performance how can we adjust the analyzer to consider the sound coming from the stage as noise and try to eliminate it from the readings? Sam: This is a really interesting question. The answer has a few parts. At high frequencies, Smaart is using a VERY short time window, so unless the stage volume leakage is coming very close to the loudspeaker, it is considered noise by the measurement. Sam (continued): At mid and low frequencies I am not sure that you want to eliminate the stage volume from the analysis. Clearly, only the part of the stage volume that is included in the reference signal you are using will correlate with what the microphone sees/hears. (Does a mic see or hear?) In this case the correlated energy is, in fact, PART of what you want to consider when equalizing. This is particularly true when in theaters that were designed to throw sound from the stage into the audience (like the Warfield Theatre in S.F.). Sam (continued): The last part of my answer deals with the time relationships between stage volume and energy from the PA. In cases where the stage energy is significantly audible in the audience it is CRITCIAL to delay the stacks so that the back-line stage volume (typically the drums and drum monitors) are aligned with the P.A. I wrote a Smaart case study about this using Lenny Kravitz’s rig that Ted Leamy formerly of Electrotec and now of JBL http://www.jblpro.com/ was tuning. Chris Engel: Sam, have you come across better ways of arraying typical 2- and 3-way boxes? Sam: Better than what?? I think that the mistake that many people make is assuming that tight packing boxes is always best. Here's a risky answer: I really like the Meyer MSL-4 as a 35-degree box. I like the sound, I like the pattern, and I think most people think it is a great box. HOWEVER, I hate them tight packed! Adding 12 to 20 degrees between boxes seems to make all the difference. Less EQ required and better pattern control. I don't know if this is how they are designed to be arrayed, but I often find that some horizontal spacing is helpful for cases like this. Sam (continued): Conversely, spacing the EAW www.eaw.com KF750’s is a BIG mistake... the coupling of the 750 horns tight packed results in great LF control... so for every rule there are several exceptions… BTW: I just had a phone call from Don Pearson, Ted Leamy and Mark Gander. They all say hi! SoundGuy: Regarding the Smaart Intelligibility Module for Smaart 3.0, I've used it in over two dozen different locations, good and bad. I've never seen a %ALCONS rating less than 2.x%, yet the intelligibility is awful (STI .55) What gives? Sam: I earlier commented that I hate those "intelligibility metrics", so here's a few more comments. I developed the Smaart intelligibility for a few reasons. One, in several European countries such measurements are required (think sales). Also, I wanted to explore the intelligibility metrics. What I found is that the equations that are used to calculate these parameters from an impulse response measurement are all a bit crazy and as I mentioned before based on values, like RT60. This means that the S/N of the measurement plays a role. Sam (continued): Further, I found that for me, the most important parameters related to intelligibility are the RATIO of the mid and low frequency decay rates, and the DIRECT TO REVERBERANT LEVEL DIFFERENCE at several frequencies. Dan: Is there a difference for tuning live sound versus for speech intelligibility in a church? Sam: Hmmm… the principles are the same. I guess the question I would ask: Is the church system full range and also asked to do music reproduction? If not, you might roll more of the lows. Also, I might spend a bit more time looking at the response with the main vocal mic on and turned up (after I had looked at it with the vocal mic turned down). Denny Strauser: Sam, how would you tune a quad system during a show? The Grateful Dead had their reference on a track that moved across left to right and back about 10 ft. This would average out the phase & room modes. Have you used this method? Sam: First the moving mic. Most people get this wrong. Don Pearson tells me that the mic stopped for measurements. I thought it was a bad idea then and now! For a quad system, you use the same principle, start with one cluster, tune each separately and then in groups. I heard the Grateful Dead "quad rig" at Madison Square Garden. I hated it! Not fade away,.. away.… away..…away.…..away.…… Chris Kathman: Jon-O of String Cheese Incident http://www.stringcheeseincident.com/ raves about you and Smaart - how did you help him use it more effectively? Sam: Jon-O is a great guy and becoming a friend! He is a very talented mixer. String Cheese is a hippie/bluegrass/electric jam band that is really fun (assuming you like that sort of thing…). When I met John-O (John O'Leary to the straight world), he was a good mixer but not thinking much about system alignment. Sam (continued): When we first met, I was working with Don Pearson at a New Year’s show, I then visited him in Boulder. We listened to the system during sound check in a 1,500-seat theater without delaying the stacks, walking from the front of the stage to the back of the house. Sam (continued): At first all you heard was stage. Then stage and stacks WAY out of time, then stacks mostly with some stage bleed. By adding a delay to the stacks to match the backline, the system was tighter and the low end punchier. The area of the audience that was hearing stage and stacks out of time (without the delay) now heard coherent sound. From there John-O has been using Smaart to tune his V-Dosc rig. His shows sound pretty damn good! I think John-O liked the fact that I emphasize LISTENING and using Smaart to help you understand what you are hearing! Roger: Since you brought it up... Some consultants have commented that they prefer Smaart for tuning but TEF http://www.gold-line.com/ for acoustics. Do you think Smaart is as capable as TEF for acoustic measurements? Sam: Actually I am extremely biased here. Whatever tools you like, you should use. However. There are some areas where the TEF provides acoustical analysis tools that Smaart does not.... the NRL noise measurement package for instance. That said... I generally do NOT like TEF. What I don't like is the ETC measurement. I think the impulse response is the critical acoustical measurement. Sam (continued): Further, I think that making measurements with a single time window is a TERRIBLE way to optimize a sound system. The BEST feature of Smaart, by far, is the FPPO transfer function, which provides a series of time windows, that result in equal resolution across the entire audio spectrum (or damn close to it). I am sure I will be burnt in effigy for that answer. Sam (continued): I would also like to add one more point: SIA was started on a VERY small budget. (My hand shook when I wrote the first $10,000 check). With support from JBL and the sale to EAW Smaart has grown and become an important tool. I hope that it will continue to grow and add LOTS of features, adding both sound system optimization and acoustical measurement features. I think most people see that Smaart has grown a lot in the past 5 years. Who knows what the future holds... Chris Kathman: Thanks for telling us so much. What is FPPO? Sam: FPPO is Fixed Point Per Octave. This refers to the ability of Smaart to provide a transfer function measurement that gives EQUAL RESOLUTION across the audio spectrum, and correlates extremely well with human hearing! Sam (continued): Smaart is an FFT-based device and the FFT’s typically provide results that are data linearly spaced in frequency, one data point every so many hertz. In FPPO mode the data points are spaces more logarithmically, much more as we hear sound. FFT is a mathematical technique to convert time domain signals into the frequency domain. The biggest trick inside of Smaart is how to create an FPPO transfer function without a dedicated DSP! Bruce N: How about the line arrays? Good/bad/ugly? Difficult to get good horizontal and vertical coverage? Sam: Ahhh - the line array question. Line arrays have been around for a long time, 70 years or more! The key issue is pattern control and power handling. Clearly the V-Dosc system is very popular and works well. JBL, EAW and many others also have line arrays. I have tuned the V-Dosc system a number of times and like it. The ability of these line arrays to control the downward lobe, getting sound off the stage is a huge benefit. Sam (continued): However, the line array is like any other speaker system in that, when used incorrectly, it can suck! All of the same rules apply; however, tuning a line array requires that you not tune each element, but rather the array as a single unit. With line arrays it is typically easier to get great vertical coverage than horizontal. Rob Scovill seems to have a number of tricks to deal with this. Sam (continued): I believe that some of the popularity of the V-Dosc www.coxaudio.com system comes from the lack of MF & LF horns. Horn-loaded systems are typically harder to tune than line arrays (in my experience). I have heard the prototype EAW rig (I was asked to join the EAW design team for a day of prototype tuning) and it sounded pretty good. I have not heard the JBL system yet but hope to soon. As for the others, time will tell. Sam (continued): I will end with this thought. Right now, loudspeaker systems all seem to provide a lot of potential, across the board, and I rarely hear systems that are really well tuned. In my experience the guys who we look up to as great mixers - the Dave Morgans, Rob Scovills, Pat Baltzells - are all guys who seem be very aware of system optimization issues, and take the time to really make sure systems are well tuned as part of their setup. I think this is a trend that will continue for a long while to come. chat.boy: On behalf of the gang at UCI/PSW, I'd like to thank Sam for sharing his time and knowledge. Sam: Please feel welcome to e-mail me at SIASOFT@AOL.COM for clarifications, questions or to send me beer! I hope you all had fun tonight.. it’s after midnight & I’m going out for a drink! chat.boy: This concludes this evening’s chat with Sam Berkow. Moderator: Thank You, Sam Sam: ‘Night all... I'm outta here! |