Apple Mac OS X Version 10.5 Leopard
Apple Mac OS X Version 10.5 Leopard [5-User Family Pack]
Pre-Order from Amazon (free shipping)
Apple Mac OS X Version 10.5 Leopard
Hal ChamberlinInterview by Brian Cowell. Way back in 1980 a book titled "Musical Applications of Microprocessors" became the companion for many Engineers designing musical instruments in the 20th century. In 1985, the book was expanded from its original 660 pages to a whooping big 800-odd pages. Till this day, the book (albeit outdated) is still seen as a source of learning for many students in this field. The man who wrote that book is Hal Chamberlin. For the last 16 years, Hal has been employed by Kurzweil (now known as Kurzweil Music Systems). There, he has been quietly working away on both hardware and software for the many products that Kurzweil make. Sonikmatter caught up with Hal (at the Young Chang factory in Korea) while he was finishing up on his design of the "remote control" for the new Kurzweil KSP8 Effects Module. Here is what he had to say: SONIK : What has been your employment history? HAL : Employment history is as follows:
SONIK : Where did you do your schooling, and what degree's did you achieve? HAL : North Carolina State University: 1973 - Master of Science: 1973 - Master of Science: Major - Electrical Engineering (Communication) Minor - Computer Science 1970 - BS Electrical Engineering SONIK : What do you do at Kurzweil R&D on a "day to day" basis? HAL : Before going to Korea full time about three years ago, I was listed as a "Senior Systems Engineer". At the time that title was assigned (somewhere around 1993). There was no standard list of titles and job descriptions so all of the senior people were just asked to suggest a title they felt was descriptive. Basically I did product development - about 2/3 hardware and 1/3 software. Analog hardware is my specialty but half of my board designs also include a microprocessor and digital circuitry. And if a micro is involved, I'm usually the one to program it. Often I'm called upon to help in systems integration and solve puzzling system problems. Perhaps the toughest such problems are the dreaded "spontaneous hard reset" which is memory loss at power-down or power-up, and audio hum. While the former usually has a "magic bullet" solution, the latter often is a tug-of-war with radio-frequency emission and safety regulations. Around 1992 I began visiting the Young Chang piano factory in Inchon, South Korea. At first it was very short trips of up to 2 weeks to oversee manufacturing startup of a new product or to solve a specific problem. Later, from about '95 - '98, it became a 3 months on 3 months off arrangement as they began to embrace computerized automatic testing of boards and complete units. By February of '98, precisely on my 50th birthday, I left Waltham for a 2-year residency that was renewed last year. I still do product development, about half the time, but now must also maintain the test systems and program them for new products. A typical day starts with a bike ride to the factory complex at 6:55AM, stopping at the company weight room (where all the piano movers train) for a 30-minute workout, then on to the Development Office where we are expected to be at our desks promptly at 8:00. After receiving e-mailed questions and issues for the day from Waltham (which has just finished its workday), I conduct an hour "class" for some new engineering recruits. The remainder of the morning is spent chasing down answers to the e-mail and often solving manufacturing problems like a sudden increase in failures of a particular production test. Lunch is at the factory cafeteria from 12:00-1:00 then the afternoon is, hopefully, free to work on whatever my current development project(s) is. Presently that is the remote front panel for the KSP8 and up until a couple of weeks ago, the new "V Series" of grand pianos featuring a PC2 sound engine. Towards the end of the day I compose a sort of newsletter ("Factory Facts"), which contains answers and solutions to the morning's e-mail plus a list of questions and issues from the factory. It arrives in Waltham about the time people there start trickling in. I usually leave about 7:00PM but often much later for the 20-minute bike ride back to the apartment the factory has provided. SONIK : How have you found the food in Korea? HAL : I'm basically an omnivorous person and so am willing to try most anything that's presented. Generally I find that the food is excellent and even the most humble of restaurants serve tasty meals. The food is definitely spicy and it's not unusual for a five-course meal to be completely red with red pepper. Kimche is really very good after a short acclimatization period. Fish and mollusks of all kinds are the primary protein sources although chicken and more recently pork and beef are becoming common. Koreans really don't distinguish among breakfast, lunch and dinner - hot fish soup for breakfast is as likely as boiled barley might be for dinner. The food is largely fat-free and healthful if one steers clear of the 3-layer pork, fried chicken in bars, and American pizza chains that are invading the country. Skim milk, however, is unheard of. Koreans are really into whole foods - whole fish in soup, a whole chicken boiled in your bowl, and whole shrimp inside a thick breading among others. However I do miss some staples that are difficult or expensive to get here - peanut butter must be imported and honey, while domestic, is quite expensive. SONIK : When you look at your career to date; what is it that you think stands out the most? HAL : For sure my 1980 "Musical Applications of Microprocessors" book is the most significant accomplishment and is directly responsible for where my career has headed since then. Actually the impetus to write it came from a friend whom I'd only met by mail. He had written a couple of technical books himself and suggested I contact his publisher. At its peak it sold about 200 copies per month, which amounted to a nice end-of-year bonus for a while. It is long out of print now after being bounced among 3 different publishers so the rights have reverted back. I still get occasional e-mail requests for photocopies. But even before that I believe I had a significant role in the development of East coast amateur computing in the 70's and especially musical applications of the same. This through (MTU), editing "The Computer Hobbyist" magazine/newsletter, writing numerous articles in other publications, and delivering quite a few lectures and demonstrations to amateur computer clubs. More recently I hope that some of my more innovative product design efforts at Kurzweil like the MidiBoard, and ExpressionMate and contributions to others have been a source of inspiration, or at least satisfaction to many. And I hope to expand that role in the future when my time is up here in Korea. Of course there are some regrets and "might have beens" too. What if I had opted to go full time with IBM instead of partnering with Dave Cox to form Technology Unlimited in 1971? Or accepted that interview with Wayne Green for editor of a "new computer magazine" which turned out to be Byte? I guess that's what makes life so interesting. SONIK : How did you become involved with the original Kurzweil Music Systems? HAL : Actually that is a pretty interesting story. As you know, I had written "Musical Applications of Microprocessors" which really covered just about everything in the world of synthesis at that time. Included was a description of and plans for a velocity sensitive computer interfaceable music keyboard. Meanwhile, a small group in Boston, called Key Concepts, had developed a keyboard instrument similar to a clavichord but with a mechanism in each key that would tighten or loosen its string as the key was slid forward or backward. It was an amplified acoustic instrument that worked pretty well but they knew that if the idea was to have any commercial future, it had to involve an electronic keyboard and be connected to a synthesizer. Sometime in 1983, I believe, one of the team members happened to get a copy of my book and upon reading the chapter about keyboards was moved to contact me through the publisher. I was offered a free trip to Boston (I was then working at MTU, in North Carolina) to see if I might be interested in working on a "radically new kind of keyboard". I took the bait and was very intrigued with the concept and excited as well at the prospect of really getting away from computers per se' for awhile and into instrument design. Anyway, in a little over a year of after-hours work, a 61-key prototype with sliding, polyphonic pressure sensitive keys was working quite well. It was attached to a Rhodes Chroma via its Apple-II interface. A slight software mod allowed the Chroma to respond to the poly-pitchbend, poly-pressure signals produced. Known as the "Notebender Keyboard", it was shown publicly a couple of times and loaned to the Berklee School of Music for a while. Lyle Mays was invited to try it for an afternoon and did some amazing stuff. The next step of course is arranging for productization, manufacturing, and marketing. After "scouring the world", Key Concepts quickly concluded that the Notebender was just a bit too radical for immediate market acceptance and besides, the Chroma seemed to be the only instrument that could use its signals. Then there was the problem of raising capital to pay for detailed design, tooling, beta prototypes, and so forth. So the sliding keys were backburnered in favor of adapting the unique pressure sensor I'd developed to a traditional wood keyboard that was already tooled and readily available from Pratt-Read. Instead of the Apple-II synth interface, which nobody else had adopted, we tried out the new "Musical Instrument Digital Interface" (MIDI), which, if nothing else, was simple to implement. While it didn't have a Poly Pitchbend command, its Poly Pressure command seemed OK for the task at hand. The result after a few more months was two "MIDIBoard" prototypes, which Key Concepts again shopped around to major synth manufacturers at the time (New England Digital, Fairlight, [what a pain it was sending their unit to Australia], and Kurzweil). All were interested this time but Kurzweil was right in town and was willing to sign a development contract with follow-on royalties. I soon found myself locked in a room at Kurzweil working frantically on the software to get a couple of units ready for the 1985 NAMM show. Concurrent with all this I had separated from MTU and after about a year of "consulting" decided to apply for a position at Kurzweil. Following a "grueling" interview, which consisted mostly of lunch with the top engineers, I was accepted and soon moved to Waltham. Besides MIDIBoard manufacturing support, my first assignment was to write a sound editor, on an Apple-II computer, for the just released K150 additive synth. SONIK : Can you tell us about the "cheesy" advert you appeared in the mid 80's that had the caption like: "our product is so great because it was designed by Hal Chamberlin"? HAL : Well, they say that 80% of advertising is getting people to remember the ad. So if you've remembered it for 16 years, I guess it was successful in that respect. On the other hand, you don't seem to remember the company or the product. The ad in question was actually run by MTU, whom I mentioned previously as my former company before joining Kurzweil, after I had left. The product was DigiSound-16, one of the very first if not the first standard product audio A-to-D and D-to-A converters for use with mini- and personal computers. Back in 1985 a major part of any serious software synthesis project was the conversion subsystem, especially if one wanted to stream data to or from the disk or other storage media. DS-16's universal (one would call it a "parallel port" nowadays) interface, and built-in 32K FIFO buffers are what set it apart. It was the last product I did at MTU and their business is still based on a long line of successor products and highly specialized software for the broadcasting industry. Anyway, I participated in a long photo session for the manual cover and computer magazine advertising and quite a lot of film was shot. Naturally the terms of departure allowed MTU free use of all materials I had developed including those photos. Why my former partner chose that particular shot (which had exceptionally bad lighting) I don't know. As for the tagline, I disavow any involvement. SONIK : At the time of writing your book, did you ever envisage where synths were going, and did they? HAL : For the first edition (1977-80) I was focusing almost completely on individual experimentation, having fun while learning, building your own synthesizer, etc. in the best Electronotes Newsletter tradition. I was not particularly aware of a synthesizer industry within the reach of poor students other than kits from companies like Paia and Emu. Essentially everything I saw that was being called a synthesizer was being sold to universities and professionals. The landscape and my experience were much different in 1984-1985 when the second edition was being prepared. Emu had really grown and New England Digital, Ensoniq, and of course the Japanese companies had quite a variety of synthesizers, keyboard and otherwise, promoted and priced for average individuals. And of course there was my direct experience with Kurzweil doing the Midiboard project. Accordingly a new chapter was created just to cover some representative commercially available synthesizers. Some really low cost technology was coming out then as well, such as the original Casio portables and the SID chip used in Commodore 64s so another new chapter about the cost-cutting tricks they used was added as well. But to answer the question, the last added chapter dealt directly with technological progress and indeed made some "bold" 10-year predictions about electronic components and synthesizers. The component technology predictions were mostly correct except for the one about DRAMs larger than 4M having to be made aboard the space shuttle to get the defect rate down. I also pretty much missed the dramatic effect that on-chip cache memory would have on future microprocessor speed. For synthesis, the prediction that sample-based instruments would be less than $1000 certainly held up as did the one that PC sound cards using sampling would become standard. I was a little off in predicting that pro level synths would separate into sound modules and controllers and that mainstream controllers would become a lot more sophisticated and expressive. Although quite a variety of alternate controllers have been introduced, we're still using piano keyboards with mono pressure (sometimes) and 2 wheels for the vast majority of instruments with no signs of that changing. Another prediction that's fallen short was that MIDI would be replaced by a LAN type technology. A number of proposals have been offered and tried during the intervening 16 years but we're still far from any general replacement for MIDI. One accurate prediction though was that all-digital studios would become common. For the next 10 years? Well, by 2012 MIDI will be replaced, synthesizers will become just virtual black boxes buried in the software architecture of the digital studio, and alternate controllers and multi-dimensional keyboards will become mainstream. SONIK : Will you be writing another book in the future Hal? HAL : I really get asked that a lot. The short answer is probably no, not as long as I'm working at Young Chang/Kurzweil because its quite difficult to get enough quiet time to do any really thoughtful writing. The first edition of MAM took about 2-1/2 years of 3 hours per night and 8-12 hours on weekends but then I was just a common engineer at a fairly large company designing high-resolution display monitors at the time. The update in 1985 took about a year while I was independent. The current schedule, which seems to keep getting tighter, just doesn't allow the necessary time to even contemplate another book. However, if I hit the lottery and retired or began working for the government, I don't think I'd try a comprehensive book again -- the field has simply grown too big and diverse for that now. Actually the most enduring interest over the years has been in the sections on musical controllers like keyboards and ribbons. Even more so, the non-mathematical treatment of fast Fourier transforms and digital signal processing continues to be popular. So perhaps a title like "Musical Applications of Sensors" or "Digital Signal Processing without the Mathaches" would be within my knowledge sphere and do well in today's market. SONIK : How do you find it now at Kurzweil as against working for the original Kurzweil Music Systems in the 80's? HAL : All companies must mature as they age and Kurzweil is no exception. I read somewhere that a large majority of Fortune 500 companies don't last beyond 20 years. Not that Kurzweil is a Fortune 500 company but we are just months away from that 20 year mark. In the beginning the company was supported by venture capital money and the overriding concern was to develop the K250 to the highest level allowed by the technology of the time essentially regardless of cost and get it shown. Although I hadn't yet joined, the fuel to power 20-hour days leading up to the first NAMM showing came not from stock options but from the thrill of creating something both groundbreaking and far beyond what had been achieved by others at that point. When I did join about 4 years later, it was time to get to work designing products that were profitable AND had a large enough market to make back the vast deficits rung up earlier. The Arnold chip and K1000 series contributed the most to this effort in the late 80s. I maintained a wall mural stock chart and it was thrilling to see the price creep up gradually before and just after the 1000PX was shown and began production. Certainly the months surrounding the bankruptcy and sale to Young Chang in 1990 were a low point but soon afterward we were running on adrenaline again as the K2000 was coming together. That was many years ago though and I'll have to admit that things had gotten to be rather routine afterwards as we crunched out product after product to keep manufacturing busy and sales up. Recently however the development, launch, and initial acclaim of the KSP8 is feeling like old times again even from halfway around the planet. SONIK : Do you think the music consumer market failed to see the idea behind the Kurzweil EXPRESSIONMATE controller? HAL : I have to give credit for the original ExpressionMate idea to Chris Martirano of Kurzweil's marketing department. Shortly after the K2500 reintroduced long ribbons to the synth market, he suggested I take that design and incorporate it into a standalone device that would let anyone add long ribbon functionality to their own synth setup. Add a few jacks for extra pedals and a breath controller and put in an arpeggiator and we'd have quick new product that could build on the ribbon technology I'd developed with our Korean membrane switch vendor. Once I'd started programming the little 6502 derivative processor inside and discovered that it was possible to play notes right on the ribbon then bend or modulate them plus some other things the K2500 didn't do, the project took on a life of its own. Which made it late, but great. Now it's really a performance instrument it its own right with three arpeggiators and an extensive MIDI processing, routing, and merging section for its two MIDI ins and outs. I think its best feature is the note playing capability. Either 1/3 or the whole length is virtually divided into 13, 37, or "scale length" little zones. Each zone is associated with a note, either along a straight ahead chromatic scale or 15 other standard scales or 8 user specified scales. Notes latched in an arpeggiator is another option. When a zone is touched, the corresponding note-on is generated. If contact is maintained and the player's finger moves away from the initial contact point, a MIDI controller, like pitch bend or modulation, is activated to modify the note. A different mode continuously tracks finger movement through the virtual zones and plays their notes as it passes through thus producing a variety of strumming and arpeggiation effects. And one can have two or even three of these going on at once for those with 3 hands. Likely the biggest problem is that the feature set outgrew the user interface. There are 135 general parameters and 16 arpeggiator parameters nearly all of which may have different values in the 3 layers. That plus 21 global parameters, most of which specify signal routing, is just too much to communicate via a 2x20 character display, a few LEDs, and 10 programming buttons. A PC based editor would certainly help and the software "hooks" are there but Windows software development is something we're really lacking in. As for marketing, I understand its failure to become a hit product. Jeff Cordero did a great job programming the "factory" setups but of necessity they had to be generalized which means that a bit of programming is necessary for any particular store display configuration. And its a fact of life now that if a new instrument doesn't do great things, right out of the box with little initial skill required, then its going to miss 90% of the market right off the bat. Perhaps if it had a built-in synth too, like the Roland D-Beam and Alesis " "air" controllers, it would have done better in stores. Still I've some great ideas for V2! SONIK : What is the story with the Kurzweil MASS chip? HAL : The MASS chip (Multimedia Advanced Sample playback System I believe) was developed in 1994-95 primarily to address the developing high-end sound card market. It was marketed to other manufacturers for use in their own non-keyboard products. If you remember, at that time a mainstream PC had a 486 processor so software synthesis was just a glimmer in a few programmers' eyes. And the typical sound card had a Yamaha FM chip or if one was lucky, an Ensoniq "wavetable" chip with a half-meg General Midi ROM. Unfortunately MASS never made it into big time sound cards or got integrated into motherboards as was hoped but the few products that did incorporate it were highly regarded. SONIK : What standard DSP chips that are currently available, do you think would be worth looking at? (eg Motorola 56000, Coldfire etc.) HAL : I really haven't kept up with the general purpose DSP chip device offerings as much as I'd like. Not too long ago a garden variety DSP would cost upwards of $100 and to take advantage of its high-speed internal ROM, one had to commit to a mask and 1,000 chips using it. Now prices are comparable to ordinary microprocessors and on-chip Flash ROM, and significant RAM makes even low volume and single's usage practical. The Motorola 56000 has always been a music industry favorite fixed-point architecture for high quality audio processing and its cost-performance has continued to be an industry leader. Its 24-bit data word length and 56-bit ALU is a perfect match to the increasingly common 24-bit audio formats. If one is comfortable programming fixed-point binary math, requires professional quality sound, and has cost constraints to deal with, it can't be beat. Texas Instruments' 32XXX line is probably the industry's broadest. At the lower end of the line are low cost 16-bit fixed-point devices selling for just a few dollars. These are an excellent choice for speech processing, MP3 reconstruction, and other audio applications that need not be pro level. At the upper end are the 32-bit floating-point models that are capable of high quality and can be programmed effectively in C - but at a cost in money and power consumption. The Motorola Coldfire is actually a general-purpose embedded microprocessor instead of a DSP chip. It is intended as a high performance upgrade for the 68000/20/40 architecture. Since it's almost completely compatible with that architecture at the assembly language level, it's a good processor that's a breath of fresh air in this era of "reduced instruction set" CPUs. SONIK : Did you do much work with VAST? HAL : I really haven't been involved in the development or programming of VAST at all. I guess the closest I got was designing the logic of the data path portion of the Hobbes chip, which executes the VAST algorithms for all of the K2XXX instruments. I have been much more involved with the infrastructure of the instruments - keyboard and player control scanners, power supplies, power amplifiers (for the digital pianos) and CPU boards for some products. SONIK : Were you surprised at how successful the SP-series stage pianos have been? HAL : Yes, quite a bit. The product idea actually originated in Korea by J.W.Chae, where he is a teacher, market researcher, and soundware engineer. Initially there was very little marketing interest in the US though because they felt that it wouldn't sell. Nevertheless, the Korean management asked me to start development and when it was completed, sharpened their pencils to where US Marketing couldn't refuse. Actually the sales surge from the SP series came at a very good time when the market for higher-end instruments was down substantially. SONIK : The KSP8 has turned out to be a lot more than "KDFX in a box"; what is it do you think that users will enjoy when using it? HAL : The KSP8 is one of those products that looks very different depending on one's approach and attitude toward it. If you're a straight-laced recording engineer looking for a subtle, accurate ambience to complement a classical recording, you'll think the KSP8 engineers had read your mind when they designed the numerous finely crafted "clean" reverbs that are available. If you approach it as a tool for enhancing and mixing more contemporary material, you'll find a wealth of delays, gates, EQs, cabinet simulators, and distortions plus combinations of these, many of which are new. And, although its not my bag, I've heard that its ability to create a space in 5.1 surround sound and place an actor in it using the joystick of the remote is just astounding. The way I approach it though is as a synthesizer. Not a music synthesizer with a keyboard but as a SOUND synthesizer. This is way beyond effects processing - its truly sound creation from source material that may be little more than random fluctuations. Many of the new algorithms, such as Degen-Regen, Super Shaper, Frequency Offset (like the old Bode frequency shifter box), Wacked Pitch LFO, Chaos and the new chaotic LFOs are reminiscent of how analog synthesis was before Moog and Carlos - just sound, marvelous sound, never before heard sound - without a keyboard in sight. Using the remote's 8 knobs and joystick to interactively control the parameters of some of these wild, feedback-based algorithms is loads of fun - just be sure your hard disk recorder is rolling. Simply put, the KSP8 is probably the most versatile piece of gear you could buy right now. SONIK : Have you ever put forward names for any of the products? HAL : Once an engineering project gets past a couple of pages and the back of an envelope, one has to name a folder to put the stuff into. Usually that changes later but sometimes the names stick. Probably the best known project that kept its initial name was the Midiboard - after all, what else would you call it, especially in 1985? - But there were others. The most internally infamous name of mine was the "RG" series of low-cost digital pianos that came out after the Mark-5 and Mark-10 high-end units around 1993-94. I had just started spending significant time in Korea and my manager there presented a challenge to design the least expensive piano possible that had the same Kurzweil sound quality but dispensed with all of the expensive features. He probably had in mind, having an acoustic piano marketing background, something with a power cord and perhaps a power switch and that's all. After it was clear he was serious, I named my design folders "Rotgut Piano". This soon morphed into "RG Piano" in order to keep some sense of dignity when discussing the project. Later as the "100" and "200" models neared completion and entered Marketing's radar screen, they decided to just keep the names never asking what the "RG" stood for. Eventually though they needed an answer for that inevitable question and came up with "Real Grand". One that didn't make it interestingly enough was "RibbonMate" which is what is still on all of the design folders and software source files. Seeing as how it grew from the original concept, Marketing decided that "ExpressionMate" was more suitable. There's another product name in the works which was arrived at much more systematically and for which I earned a bottle of Korea's finest traditional liquor, but I can't talk about it yet. SONIK : I hear you still sell Kurzweil K150's? HAL : Yes, but only around Christmas time when I'm in the 'States' for a few days. - US$250 for an "old style" unit with switching power supply. Ed note: You can email him at: Hal Chamberlin. SONIK : Have you looked at some of the other keyboards/modules that used additive synthesis? HAL : Besides the Kurzweil K150, the only other additive synthesis instrument I've actually used is the Kawai Kawai K5m. It too is quite old and really doesn't do classical additive synthesis being that partials are locked together in 4 groups with a common amplitude envelope and must be harmonically related. The later K5000 was an incremental improvement but I haven't tried it. I've also heard of, if not heard, several others including early New England Digital, Synergy, the "GDS" (General Development System), and the Technos Axcel. The latter had what amounted to a very large, lighted, but coarse resolution touchscreen where one could "draw" spectra in real time or interact with analyzed spectra while a sound was playing. I don't know of any recent hardware that uses primarily additive synthesis other than some Hammond B3 organ emulation modules by Voce' and others. Perhaps when the current retro analog craze runs out of steam and processing power cheapens even further we'll see a resurgence of interest in additive techniques. Naturally there is a lot of software around that does additive synthesis and "resynthesis" using fast Fourier transforms but I have not tried any of it. Also a web search reveals that additive synthesis is still a popular thesis research topic, which bodes well for the future. SONIK : What forms of synthesis would you like to see being explored more? HAL : I think that pure, or nearly so, additive synthesis needs to be revisited commercially. When the K150, K5, and other early models were designed, 256K RAM and 8MHz 286 processors were the norm in computers. Now that processors are 200+ times faster and memory is 500 to 1,000 times cheaper, one should be able to make an interesting and saleable additive instrument. Physical modeling too has endless potential but mostly in directions different from what Yamaha was pursuing. Perhaps just "modeling" would be a better term because the most interesting models are likely not physical at all. What made FM synthesis, which after all is a model albeit a simple one, so interesting was the surprises one would encounter when working with it. Additive, which is really modeling too, can sometimes be surprising as well, especially when dealing with it through meta controls rather than one partial at a time. No doubt there are other models with the surprise and richness of FM combined with the generality of additive just waiting to be discovered. Nothing, however, is going to replace sampling for mass market instruments because, given cheap memory, it is simply the most cost-effective way to provide a lot of realistic instrument sounds. SONIK : What is your view of the "hardware versus software" debate? HAL : Hmmm, that's a tough one because there are so many different ways to look at it. Of course at their heart, the digital computations, "hardware" and "software" techniques are identical. Any result achieved by one could, in theory, be achieved by the other. I'll try to touch on a few of the issues but after just 5 minutes of thinking about them, its clear that one could write a good sized book on the subject - that would be out of date in a year. Anyway, consider the kind of PC or Mac software-based system common today. One has a large number of essentially independent hardware subsystems like the CPU, main memory controller, cache controller, disk drive(s), A-to-D and D-to-A converters, perhaps a MIDI interface, and so on, all made by different manufacturers. Even more numerous are scores of software components, including those devoted to synthesis and sound processing, that were all written independently. The key is that each of these components, both hardware and software, takes a variable amount of time to do its task depending on circumstance and what other components are doing. There can be external influences too like interrupts from the network interface or mouse movements or something that triggers a window redraw. The operating system coordinates all this to produce an accurate final result but the path and time taken to do so can vary tremendously. An analogy to the software system might be a construction site with individual workers, trucks and other equipment, and outside forces like deliveries and weather, all working independently. At the same time they are seriously interdependent for getting materials, sequencing tasks, and so forth. The time to finish the building is statistical with an average and (typically large) variance. One way that software synthesis deals with the uncertainty is through the use of buffers between components. If an upstream component is held up, for example, it got interrupted for a screen redraw, the buffer following it can still supply data to the next component downstream. If that buffer should run out, then the next buffer in the chain keeps the data flowing. If all of the buffers run out, then one hears a click or program interruption. Larger buffers allow for greater variance in processing times before defects are heard but at the expense of a greater delay (latency) through the sequence of steps. Another way to cut down on variance in a processing step is to utilize only a fraction of its theoretical power. A synthesis routine for example might have data structures defined for 100 voices. For all to be active simultaneously, it might need, say, 70% of available CPU cycles and 80% of available memory access bandwidth. Almost any event in the system would disrupt this for a time thus requiring the following buffer to take over and after the disruption, the buffer would refill slowly. However if only 50 were playing, larger disruptions could be tolerated and equally important, the buffer would more quickly refill afterward. The performance of software-based systems can be improved most effectively by using an operating system specifically designed for audio processing instead of a general purpose one like Linux or Mac OS or Windows XXX. Since that's not an option at the moment, elimination of extraneous tasks (you don't really need a stock ticker running while playing do you?) is probably next most effective followed by using the fastest processor and largest memory one can afford. Another issue in the debate regards hardware vs software sound quality. Software synthesis still must take a lot of computational shortcuts in order to produce a reasonable number of voices on present PCs and this adversely affects sound quality. On the other hand, applications like recording or unaltered sample playback or mixing, which are really just data movement, won't suffer from computational shortcuts and so can sound excellent. Computers make wonderful recording systems! I'm sure everybody has noticed that the number and rate of introduction of software synthesis products is many times that of hardware. Indeed it seems like hardware based development has slowed way down from, say, the late 80s and early 90s. Mostly this is due to the great and increasing capital resources needed to develop a new hardware-based synthesizer or signal processor. Custom VLSI, which has always been very costly to develop, is several-fold higher now than it was 10 years ago. Much of that is the far higher cost for .25 micron and under masks compared to 1.5 and 2.0 micron then. But a large part also is competing against cell phone, Internet equipment, and PC manufacturers for chip manufacture in the thousands when they are demanding it in the millions. Aside from chips, the regulatory burden is higher now and tooling costs for PC boards and enclosures are up. Often the synth manufacturer is now forced to use components that are great for cell phones or PCs but are less than optimal for synths. So it's really tough and getting tougher to design and build keyboards, racks, signal processors and other hardware based gear. Now consider the relative ease of developing software synthesis and signal processing applications running under a general purpose OS. Unless you're out to solve the world's problems, most all of the infrastructure - like sample reading and writing, buffering, A-to-D and D-to-A conversion, and GUI elements - is already there and adequately functional. One merely needs to dig up the documentation and start coding away on the core algorithms of interest. And even that is much easier now with high-speed floating-point hardware being a standard part of the processor. With floating-point arithmetic, programming in C becomes nearly as effective as assembly language. No longer is highly specialized knowledge of the mathematical properties of binary arithmetic necessary to get a good performing DSP algorithm running. Producing software has always been much easier than hardware, and with the Internet, it is getting easier. The net result is greatly increased accessibility and democratization. With so many "brains" working on software synthesis and other DSP applications, a lot of good stuff (along with plenty of junk too) is coming out now. I am especially amazed at the progress heard at last year's NAMM show in real-time software-based time/pitch shifting. The future? There will always be hardware and low-end mass-market synths and digital pianos will continue to use custom chips and small microcontrollers running highly optimized dedicated software. As one moves up, there will be increasing use of standard DSP chips in place of custom VLSI and general-purpose processors controlling them. The high-end "hardware" synth of 5-10 years from now may be little more than PC hardware plus a really good audio interface packaged in a typical keyboard enclosure. The casual user will never realize this however. SONIK : What activities do you do outside of work Hal? HAL : Excluding personal electronics, programming, and musical activities after hours, I'd have to say that number one is definitely bicycling - especially long-distance touring. I'd been riding around town (in preference to driving) since the late 70's. But about 15 years ago John Teele at Kurzweil introduced me to self-contained bicycle touring and camping. Since then I've crossed the US West to East twice, in 1994 and again last summer. One of the perks from spending time in Korea is accelerated vacation time accrual, which made these 3-month journeys possible. In Korea, the traffic and other problems make bicycle touring less inviting but most all of the country is mountainous so day-long and occasionally overnight hiking and climbing trips with co-workers are eagerly participated in. And the Kurzweil Bowling Club, later replaced by the Kurzweil Racquet Club, are some other activities I enjoy, if not excel, in. SONIK : What instruments do you have in your own personal studio? HAL : Sadly, I've been in Korea away from my primary setup in Waltham for a bit over 3 years now. Before putting most everything in storage, the music/computer room had the following:-
For the Korea apartment, I managed to haul along one of the K150FSs, the Apple-II, the Alesis items, and the MIDI router. To that has been added a K2500R, SP88X, and dual ExpressionMates. I also bought a new PC and enjoy a 1Mbps DSL Internet connection. Since sound levels have to be kept low in the apartment, the "monitors" are basically the sound system from a Kurzweil RG100. I think you can see that I have a special fondness for the K150FS and indeed true synthesis of all types. I really like to start from nothing when making a sound rather than modifying something existent. I mostly putter around and don't do much with sequencers; waiting for that large block of time necessary to undertake something serious. SONIK : Who have been your musical influences? HAL : I have always liked Bach fugues and classical organ music in general since being a little kid. I once got into trouble climbing around the pipe room at my church taking pictures and examining how the various parts worked. I never really embraced pop music although there were a number of tunes from the 50's and 60's by the Beach Boys, Del Shannon, and similar groups I could relate to. After getting a BS in 1970 though I basically tuned out nearly all pop music, which had become too disordered and loud for my taste. There were exceptions though, like the Moody Blues and The Who. I'm probably not alone in revealing that Switched On Bach by (then) Walter Carlos had a substantial influence. Before that one might say I used my rig primarily to scare the neighbors and amuse the cat, but afterward I saw that my hobby might actually have some career potential outside a B movie set. More recently I'm into Vangelis, Toto, Yanni and a few other more obscure "new age" composers. My all-time favorite album is probably the soundtrack for Dune. And some of the Korean classical and even pop I'm bathed in every day now isn't bad! SONIK : What would be the funniest/strangest things that you have seen in the music industry? HAL : The weirdest experience at Kurzweil I can recall happened late at night around 1993 when I was trying to find the source of a low, but audible in a totally quiet room, hum in a prototype Mark-10 piano. Another engineer and I had been swapping boards and cables and trying various grounding experiments most of the evening without a whole lot of luck. After making a small change to the MIDI I/O board, we turned the unit on and after about a minute it started playing itself! What we heard was eerie but very musical and we sat listening for several minutes. After much speculation, we concluded there was a bug in the software (which is natural for hardware folks) that caused it to play portions of the demonstration songs at random. But on further listening that really didn't fit either. Eventually we discovered that a wire in the MIDI in cable had opened and was "floating" thus feeding semi-random bits to the MIDI interface. I really wish I had a recording of what came out of that unit that evening. SONIK : You have a unique perspective of intellectual property issues for musical digital signal processing, because your book predates the granting of many patents in the field which are considered by some to be controversial. In general, do you think the US Patent Office does a good job in granting musical digital signal processing patents? HAL : First, I am not a lawyer, certainly not a patent lawyer. That said, my opinion is that it is far too easy to get a patent. Patents should not be granted to practitioners who are merely doing their jobs unless that work product shows extraordinary insight or creativity. Patents are routinely granted nowadays for the solutions to problems so obvious that students in an engineering course would be expected to arrive at a similar solution had the problem been posed as a homework assignment. So much of what passes for innovation nowadays is just obvious applications of available technology. Ten or 20 years ago, before the enabling technology was common and cost-effective, those same ideas might have been innovative and worthy of a patent but not now. Remember, a patent grants one the right to prevent anyone else from arriving at the same solution and using it in their own work without permission or payment for 20 years. I think that a piece of work needs to be held to very high standards to enjoy such a degree of protection. Hopefully standards can be tightened in the future, but it would have to be a gradual process to be acceptable to those accustomed to routine granting of patents for routine work. Probably more important than the book you mention (which I began writing in 1977 and was published in early 1980) was my Master's thesis at North Carolina State University. It was titled "Design and Simulation of a Digital Sound Synthesizer" and was placed in the NCSU library in the Spring of 1973. It basically described how a modular analog emulation synthesizer of 16 oscillators/filters/amplifiers, effects, and routing array using memory IC's could be put together with existing technology in quite a bit of detail. I had planned to build the device but my thesis advisor suggested simulating it instead on the campus Ambilog-200 computer. Later, as the deadline approached and I was pulling all-nighters just getting the simulation working, I was thankful for the advice. I even got some actual music out of the thing; a reasonably listenable Bach "Toccata and Fugue in D-Minor" which now exists only as an analog recording and a roll of punched paper tape source code. Anyway, that thesis, in conjunction with a lot of other materials, was instrumental in settling a recent patent suit against Kurzweil (which I cannot discuss) that might have spread to the entire sampler and wavetable sound card industry had it been lost. SONIK : Does the patent examiner's decision reflect a knowledge of prior art, and a good judgement of what is non-obvious to a skilled expert in the field, in your view? HAL : In my experience, the examiners are pretty good in going through existing patents and finding most if not all of the prior _patented_ inventions or solutions (called "art" in patentese). They are generally poor at finding, or even searching for prior art in books, magazine articles, theses, conference proceedings, and so forth. This may be improving with the Internet making it easier to search such materials but there is still great reluctance to disallow a patent claim unless it is exactly the same as something done or published previously. SONIK : What advice would you pass onto Engineers looking to start a career in the music manufacturing industry? HAL : I guess you mean electronic musical instrument industry. One should probably start by obtaining a thorough background in musical acoustics - pitch, amplitude, timbre, waves, etc. then supplement that with some study of high fidelity sound - frequency response, distortion, hum & noise, something about loudspeakers, etc., and finally cap that with at least a course or two in music theory. Some ear training and minimal skill in playing an instrument or two will be very helpful if your work will involve judgement of sound quality or instrument performance. I think these are fundamental for any "musical engineer". Beyond that, if one is interested in synthesis engines, solid coursework in mathematics, linear systems theory, and of course computer architecture and some programming is a must. If the focus is toward controllers and player interfaces to instruments, I'd recommend substantial study in analog circuits, electronic instrumentation and measurement, and sensors. And a course or two in physics of materials wouldn't hurt either. "Hobby work" is important too. If one is not interested enough in instrument design to have spent and continue to spend free time on personal instrument projects, a good job may still be possible but long-term creative excellence is likely to be elusive. Finally, unless you're going to work for one of the big 4 or 5 companies in this industry, its crucial to understand economics in general and business economics in particular. They may call economics "the dismal science" but that's only because if an entrepreneur doesn't understand it, the result will be dismal. At least one course in basic business principles and perhaps one in marketing may well be as important to one's success as all the previously mentioned stuff. SONIK : If you were to move out of the music industry; What field would you like to work in? HAL : Robotics is intriguing and before moving from North Carolina in 1986 I was an active member of an amateur robotics club there. Accurately and gracefully controlling motors and coordinating multiple movements is really a lot like sound and music synthesis. It's certainly a much bigger market and also more focussed on useful results instead of style which sort of fits my mindset. Another area of interest is the whole field of energy generation, efficiency, and conservation. One of my contributions to the Kurzweil technology base has been higher efficiency linear power supplies that produce less heat without switching noise. Doing more with less - of any resource really - has always been an enjoyable challenge. Yet another possibility I've given more than passing thought to is starting a small electronic gadget company, perhaps with initial emphasis on bicycle accessories. Over the years I've developed quite a laundry list of possible cool products that address specific problems but have never really been able to pursue any of them. Might be a good "retirement" pursuit. Additional Links: Jan 2002 |
|