Episodes of the Mid-O Series

Timeless truth is that the makers invented instruments, whereas the artists reinvented them into musical instruments. (Citation needed).

The Beginning

In 1983, a number of instruments were released with MIDI, the standard that I was involved in. After the tough times with all the difficulties of development, Roland launched the TR-909 the successor model of TR-808. Yamaha made debut of the ground breaking DX7, a serious digital synthesizer and again with MIDI. It was the opening of the new chapter in the electronic musical instruments, the dawn of full digital synthesis. 

Roland then was still a new emerging enterprise, and did not have enough resources to pursue digital technology. To combat the situation, the Fundamental Research and Development Department was established. 

Transferred to this dept., was the product R&D staffs previously responsible for the Mid-O Series models like the DR-55, TR-808, TR-606, TB-303, and TR-909. Their new role was to concentrate on digitalizing electronic musical instruments. 

Because of this, we were somewhat detached from the actual product R&D since then, hence didn’t have much access to the market trend. It was not until in 1992, when I visited Tokyo Sales Office, that I learned the TB-303, which we had no memories other than an utter failure in terms of sales, was now huge in the underground music scenes of Europe and US. 

Since then, there seems to be numerous inquiries coming to the company from many media and journalists around the world about how we developed the models, not just the 303, but also the 808, 909 etc. Some of them were directly sent to myself, and I always replied to them as follows; 

“Roland Corp. is happy to see the Mid-O Series being popular and are highly regarded in hip-hop, techno, and dance music genres. However, these instruments have evolved into something far beyond our original intention and expected usage, and even value. We are very much obliged to and at the same time overwhelmed with your kindness. To express our gratitude, I would like to quote old words that goes like this; 

`Timeless truth is that the makers invented instruments, whereas the artists reinvented them into musical instruments`. “ 

Many of the interviewers seem to be satisfied with this reply, but there were still some who asked deeper into the R&D episodes. I was sorry to say that I had no other choice but remain silent on this.

Almost 40 years have passed since then, and we are so much grateful to see the sounds of Mid-O Series now regarded as vintage being heavily used in diverse musical genres of today. Also, after the Mid-O Series, we delved into digitalization of musical instruments, establishing our new foundation with technology for the RD-1000 and D-50, then forging into the future with V-Series such as V-Guitar, V-Bass, V-Drums, V-Synth, and V-Piano. We are so happy to be able to get involved into fundamental technological development for these models. 

Taking this opportunity, as the respect to the artists who turned our offspring the Mid-O Series into masterpieces, I would like to introduce you the episodes of the original intention on developing them, and technological background as well. 

Also, it is noteworthy to mention that the aim of the TR-808 development team was to design not a drum machine, but a drum synthesizer. It was not for mimicking a real sound, but to provide possibilities to create ideal sound. However due to the limited performance, the physical volume, and the cost of the analog devices at that time, we had to abandon this original target that we were pursuing.

Then, in this past 40 years, the PC processing power increased so much that now it is possible to have DSP and GUI to virtually reconstruct an entire set of Roland modular synthesizer System-700 to generate a single percussion instrument sound. Of course, nowadays instead of synthesizing a sound you can get abundant number of realistic samples of various sounds. The TR-808 being no exception, and because it is regarded as a vintage instrument, it is deeply sampled and those samples are vastly used. But then again, like paintings are with the photography, synthesizers are with samplers. To show the meaning of a drum synth, I decided to recreate it and challenge once again to design it with contemporary technology as the RC-808. Note that it is my aim to develop a system without PCM sampling, nor digital effects computing such as delay, reverb, chorus, etc., but according to the analog processing style of at that time. In other words, this is not an analog modeling, but an analog manner synthesis. 

The synthesis is analog manner, nonetheless synthesizing percussions are far more complicated than classic analog synthesis sounds. Required for acoustic analysis, and synthesis are the fundamental theories and knowledges of signal processing, and acoustic technology. Sadly, my programming techniques are not enough to eliminate some bugs, nor minimize latency and jitter. Moreover, there are not enough documents nor manuals that should be attached to the synthesizer sound sources, nor acoustic analysis tools necessary for synthesizing a sound. Furthermore, I have not started to develop the macro controller that allows to collectively control the numerous partials of a sound.

Therefore, the RC-808 is still under development. I and my team here cannot bear responsibility whatsoever of any possible inconveniences or losses caused by the RC-808. We cannot answer to inquiries for the RC-808 as well. Because the RC-808 is supplied as a conceptual model, and as a freeware that can be down loaded freely, it is alright for us to have someone else use it for creating new drum sound. This might realize our goal that we aimed some 40 years ago.

The “Sequences“ to Develop the TR-808

In 1979, returning from his business trip to US, president Kakehashi told a following story to me who was a chief engineer of drum machine development at that time. “There is a demand for realistic sounding drum machines from the recording studios in US. They wouldn’t have to hire drummers, so they can lower the cost of music production. If we were able to provide it in below a thousand dollars price range, then there will be a huge market. 

Perhaps he brought me this story, because we already succeeded in developing the BOSS DR-55 Dr. Rhythm. It sounded worth to challenge, so I undertook it, perhaps a bit too adventurously, and initiated a new R&D project at once. Kakehashi-san didn’t mention the source nor further information, but several months later he handed me a cassette tape he received in US. It was a recording of a newly announced American digital sampling drum machine called LM-1. People were seeing the new possibilities to apply the PCM sampling to electronic musical instruments with great expectations. At the same time, it was hefty costly, and in fact the LM-1 was with the price tag of 5,000 US dollars.

Roland rhythm boxes were based around analog technologies made possible by engineers like M. Shibahara from former Ace Tone, and T. Sakai. This technology was then modified by T. Kamio and H. Nakamura to be deployed on the CR-68 / -78 series.

The Ace Electronics Industry, known for its Ace Tone brand, supplied rhythm machine as an OEM manufacturer to Hammond. At NAMM Show, Don Lewis was playing this rhythm machine in Hammond booth, and this particular unit he was using had been modified to have an expression pedal being directly connected to it. Being surprised to hear the sounds that are unexpectedly different from his own manufacturing company product, Kakehashi-san started talking to Don. This episode was widely known within Roland as well. It would be great if you could also refer to the R&D episodes told by Don for the CR-68 / -78 models. The drum machines that followed the DR-55 were designed to be complete with the Accent function. Also, it could have been Don who brought the LM-1 news first to Kakehashi-san while he was on a business trip to the USA. 

The DR-55 has the analog circuits that are simpler than the CR-78, but again it sounded far from realistic, more like a toy so to speak. On the other hand, the LM-1, although not hi-fi yet, realized the essence of the real percussions packed in extremely short sampling time.

We calculated what the price would be if we developed ours with PCM sampling technology, but then again even the kick drum would require several hundred milliseconds, and cymbals several seconds long sonic duration. With the semiconductor memory cost at that time, the estimated price tag of our product would be over ten thousand US dollars. However, what made Linn great was the spirit to tackle the difficulties by boldly giving it a try, and the superb skill to replicate the kick drum sound with mere 150 millisecond sample with only 8bit resolution.

The PCM sampling was expected to become the next generation mainstream of musical instrument technology, but we synthesizer engineers couldn’t stop feeling that simply recording and playing back the acoustic sound was too easy, and is not for someone with craftsmanship. 

Therefore, because Roland is a synthesizer maker, I suggested that we should go for developing a drum synthesizer with each instrument having its own synthesizer that allows you to create the sounds of your taste, rather than just realistic sounds. Kakehashi-san agreed with this idea.

Before I was hired to Roland in 1977, starting from around 1972 I was developing multiple systems with MPU such as Intel i8008 and so on. Even before that, I was an expert of linear circuits such as transistor radio, TV, wireless systems etc. After stepping into the electronic musical instruments fields, I was surprised to see this unique technological culture that boldly and cleverly utilizes the non-linear domain of semiconductors that was considered as taboo in the linear system. This non-linear technology enables the distinctively analog sounds, but at the same time, it had large individual variance, and lacked the dynamic range, S/N, temperature dependency, and stability of the products.

At that time, we made countless tackles to create realistic sounds with Roland flagship model System-700 the large-scale modular synthesizer, which were tough challenges indeed. In order to set the price under a thousand US dollars, we had to limit the electronic components both in kinds and quantity to be used. We also had specification limits with dynamics and stabilities of semiconductors. The biggest problem was that deploying numerous potentiometers to allows users to synthesize sounds will consume so much space on circuit boards that we will quickly run out of rooms to place electronic components that comprise the electronic circuitries. This greatly forced us to lose the freedom on designing the product.

Observing the attack portions of percussion instruments with oscilloscope showed us extremely complicated harmonic activities going on. We found that this active harmonic changes occurring in a short period of time was essentially producing most of the identity to each kind of musical instruments. The analog subtractive synthesis invented by Dr Moog expressed the attack portion with only the rise time of the envelope generator and the cut off frequency of the VCF, which was very practical however also with certain limitation in sound variations to be synthesized. Hence, we tried to develop the product as simple and low cost as possible but also tried to design dedicated circuitries that enables to generate particular characteristic attack portions unique to each percussion instruments.

Bass Drum

It is the most important percussion in all music genres. As mentioned above, there is an extremely complicated harmonic activity in the first 50 milliseconds of the attack portion. Furthermore, common with the other drum sounds are the numerous elements ranging from several hundred hertz to several kilohertz in the transient segment. Moreover, they are high-pitched at first, then rapidly but smoothly goes down to lower frequencies, and then gradually decay into the fundamental frequency of the drums that is 60 hertz. Portions of these our discovery was later turned out to be known already in other fields as well. For example, in the world of DSP technical terms, the signal that changes from high to low frequencies is called as Down Chirp, Swept Sine, or Time Stretched Pulse, and is applied to radars and measurement equipment. We tried to enhance this method further with the successor model TR-909, but it was drowned by waves of sample technology drum sounds. Then again, I heard that this synthesis skill was analyzed and applied to circuit bending skills for the 808 / 909 clones in both analog methods and software also.It is the most important percussion in all music genres. As mentioned above, there is an extremely complicated harmonic activity in the first 50 milliseconds of the attack portion. Furthermore, common with the other drum sounds are the numerous elements ranging from several hundred hertz to several kilohertz in the transient segment. Moreover, they are high-pitched at first, then rapidly but smoothly goes down to lower frequencies, and then gradually decay into the fundamental frequency of the drums that is 60 hertz. Portions of these our discovery was later turned out to be known already in other fields as well. For example, in the world of DSP technical terms, the signal that changes from high to low frequencies is called as Down Chirp, Swept Sine, or Time Stretched Pulse, and is applied to radars and measurement equipment. We tried to enhance this method further with the successor model TR-909, but it was drowned by waves of sample technology drum sounds. Then again, I heard that this synthesis skill was analyzed and applied to circuit bending skills for the 808 / 909 clones in both analog methods and software also. 


To meet the low target price, I employed oscillation technique of pulse wave modulating the simple T-network bandpass filter. By short circuiting the resistors in the T-type network for the certain initial duration with transistors, it was possible to raise the frequency. This allowed us to enforce the attack sound compared with conventional rhythm boxes. 

But the initial reputation of the 808 was nothing but crippled as it was compared with the Linn products. Furthermore, the sound had insufficient impact or punch at the beginning, which resulted to have pronounced fundamental frequency 60 Hz. But then again, it was this prominent fundamental which attracted the attention of later time hip-hop and EDM artists, and they gave a name to it as “Deep Decay”. This led to the rediscovery of Mid-O Series. The long decay sound was made possible by decay parameter that I added to show at least as a sign or as an evidence of the 808 being a synthesizer. Even inside the company this was regarded as going too far, but later it became one of vital identities of the 808. Nowadays you can hear pretty much of its sound even with small monitors or even with headphones, but back in the R&D day of the 808, we had only cheap playback devices in our labs in Roland, and none of us were able to hear nor realize the heavy low bass frequency bands of this killer tone.

Handclap

The LM-1 had a handclap sound that was not present in the preceding analog drum machines. It was digitally sampled but the sound itself was muffled due to infant technology of at that time. We actually clapped our hands in a room with live ambient acoustic reverberation being heard, and observed the audio signal from the microphone with an oscilloscope. My analysis revealed a random signal sound source with central frequency of approximately 1,000 Hz, which was then passed through a series of amplitude modulation comprised of several sawtooth shaped envelope. Reasonable outcomes were yielded by passing the white noise through a bandpass filter tuned to 1,000 Hz, and then amplitude modulating it as mentioned above. Then again, it didn’t sound like popping enough, but more like a bunch of bamboo sticks being hit, however we were running out of time. After we released the TR-808, I succeeded to improve it to a better sounding circuitry complete with popping sound generation, but the 808 handclap sound was already gaining a lot of popularity, and it was too late to replace it with the new circuit, nor launch the new one. This handclap sound is the de facto standard in the market. And together with the kick drum sound, it became the identity of the 808.

Just before the 808 was officially launched, the sales dept seems to have rented it to YMO, so I was surprised to hear its sound from an FM radio broadcasting program of their live concert in Budo-kan Tokyo. 

Cymbal, Hi-Hats, Cowbell

Previous electronic drum metal percussion sounds were colored noise that were achieved by white noise being passed through high-pass or band-pass filter. However, the actual sounds were not smooth enough, nor had metallic timbre. As a countermeasure, it was already widely suggested that combining a number of pulse generators would provide better results. The metallic noise obtained by combining six pulse wave generators had spectrum energy being scattered and collected in relatively favorable manner. The unwanted low frequency elements will be eliminated by filter. The resulting signal already sounded metallic but because it is from differentiation of pulse signal, overlapping multiple generator outputs will bring prominent amplitude and the resulting sound will be harsh. Molding the signal with envelope from low cost chopper type VCA will lower the prominence into more average values but at the same time brings sidebands at the either side of the spectrum. The noise that appears as the lower sideband has unpleasant dirt, and eliminating them with low cut filter produces not realistic but silky metallic noise unique to the 808. These metallic sounds owe largely to experiences and efforts of H. Nakamura who was involved in developing the CR-68 / 78 sound engines.

Snare Drum and other percussions

Snare drums are the second important drum instrument in the drum machine next to the kick drums. However due to a lot of time being consumed for developing the above-mentioned bass drum, metals, and handclap, we ran out of time when we made some modification to the CR-78 snare drum circuit for the 808. We already had schedule delay longer than expected, and the circuitries became larger in scale also. In order to squeeze them into a small cabinet, clever circuit designing and PCB skills were required. This was made possible mostly from contribution of K. Fujiwara who was a freshman at that time.

Multiple Outputs

As soon as the prototype TR-808 began to work, Kakehashi-san brought many musicians to the lab and asked for evaluations. The multi-outs were already present on LM-1, so we followed to their advice. As described later, the new visual method of rhythm programming system was highly regarded. Then again there were those who compare the sounds with LM-1 that had realistic sounds supplied by sampling technology.

Sequencer

The CR-78 programmable rhythm machine had programming capability by tapping the button. On the contrary the low-priced model BOSS DR-55 Dr Rhythm had non-realtime typing method. Some of us in the company did concern about it being too machine-like, but it turned out to be well favored. At that time, Roland had a gigantic 24 step x 16 instruments matrix switch board that was used for programming the rhythm patterns. There was an idea of utilizing microcomputers to integrate this huge board into a mere single row of switches and light-emitting diodes that are placed along the individual steps of a timeline.

We believe that this rhythm programming system that allows to input visually and without hassle owed greatly to the resurgence of the TR-808 in the underground music scenes of Europe and US, after it disappeared from the market as it was miserably regarded as cheap sounding compared with the LM-1.

The now iconic red, orange, and yellow tap switches are same as the ones that are used on the flagship model of Roland at that time, the 8-voice synthesizer JP-8. 

The origin of the model name TR-808

There were also many who asked for the reason why the TR-808 was named so. The musical instruments industry of the ‘70s when Roland founded was still filled with the buzzword saying the transition from vacuum tube to transistors. Curiously enough the new computer age was already happening, but it was decided to carry on the name of transistor rhythm TR to the new model. Then again, Kakehashi-san put zero in the middle to differentiate it from previous models, and thus the new Mid-O Series like the TR-808, 606, 909, TB-303, MC-202, SH-101 was launched.

The “Sequences“ to Develop the TR-909

As mentioned above, developing a drum synthesizer went into direction far from the original goal due to the limits and cost of analog components. But after developing the 606, 303, and 202, while the age of sampling was coming fast with big strides to the musical instruments industry, once again we had a chance to challenge developing another analog drum synthesizer. By then the price of OP-Amps were becoming low, so we aimed to use them for aforementioned Down Chirp Oscillation to realize dynamic bass drum, snare drum, and improved metallic sounds with pulse generators.

However again, although with large cost drop and bigger rooms to place components available, the synthesized sound variation was far from satisfactory. As the time was up, we were forced to introduce PCM sampling technology for metal percussion sounds such as cymbal. We already developed a four-voice sampler inside Roland, but then A. Hoshiai who was in charge of metallic sounds for the 909 developed a clever and practical sampling method by using a low cost 6-bit memory like ADPCM. This highly skilled method was succeeded in developing the next drum machine with PCM sampling technology the TR-707. Also, the TR-909 was world’s first drum machine with MIDI.

Kakehashi-san, being worried about the R&D delay of the 909, came to our lab and advised us to switch to PCM sampling on drums as well. After extensive discussions with Y. O-e who was in charge of designing the sound source, I gave up the idea of implementing a universal synthesizer to the 909. Instead, we pursued making attack with impact that were not available on the TR-808, and this evolved into a compressor-like sound. We had Kakehashi-san listen to the sound and obtained acknowledgement on continuing the development, but here again the TR-909 turned out to be a failed product. There are no words for us to express our gratitude and apology to late Kakehashi-san, however this failure eventually became the basis of next Roland digital technology, SA digital pianos, and LA synthesizer.

The “Sequences“ to Develop the TB-303

The TB-303 was developed to depreciate the molds of the body case and knobs for the TR-606, which came after the TR-808 as a lower priced model, and at the same time as a programmable bass line that runs in sync with the 606. The only available low-cost display was the LED, hence unlike the visual and friendly programming on the 606, the 303 system became quite difficult and tedious. Ironically this tediousness invited operation mistakes that brought unexpected patterns and sounds by accidents. Computer programming was made by H. Matsuoka who had been in charge since the 808, but it became the most painful product for him to develop. The sound source was developed by Y. O-e, but again, although it was aimed to provide realistic bass sounds, the actual outcome was utterly different. The VCF developed here had a cleverly designed stable feedback circuit so that it will not self-oscillate, and perhaps this led to the bold sound designing in the later Acid music etc. Here again, it was the hungry and aggressive artists, who made this failed instrument into a masterpiece.


RC-808

An artist, being asked the reason why about using the TR-808, replied “Because it is a standard”. This does not mean that it is a nostalgic popular sound that has been used since long time ago. In the world of aesthetics, not just music, the sense of beauty and value is essentially comparative evaluations with criterion. I consider this criterion as not a goal or singularity within timbre space but as a vector that has a force and a direction. They may have found the TR-808 in a corner of a junk shop in the early 1980s when the electronic musical instruments began to output realistic sounds through sampling technology. And out from that junkpile, they discovered a new criterion, and created a completely new kind of music. Likewise, the TB-303 and the TR-909 became vintage instruments in the new world of music created by this new criterion.

Katsushika-Hokusai

If you liken it in the picture art world, the TR-808 sounds are simple line drawings and illustrations so to speak. It puts more weight on showing essence simply rather than realistic photography. The traditional painting of Japan known as Ukiyo-e is a pictorial method and manner that is primarily made of outlines derived from specification of printing and paints. This manner has its own limits, and hence is a vector of criterion. Katsushika-Hokusai, famous for his Mr Fuji pictures, created The Great Wave off Kanagawa(神奈川南沖浪浦 Kanagawa Minami Oki Nami Ura). In that piece of work, he threw off the photographic details into abstraction that boldly enhanced the great waves by outlines, and made contrast with a pointed Mt Fuji. This manner derived from its own limits is something impossible with realistic paintings that put priorities on details. 

The inventor of the bass drum must have done trial and errors like enlarging the diameter or making the membrane thicker to produce fat low sound. Large portion of the beater strike energy to the skin however must have been consumed by uneven fractional vibration or parasite vibrations, resulting unwanted muddy sound. Also, the drummer spends a long time doing a lot of work on damping with blankets and else in order to suppress the unnecessary resonance with the body or with other drums. But once adequately used, this dirt will become important part of the sound, its identity, reality, and criterion. 

The TR-808 bass drum was without the reality of the drums but was like an enhanced great wave depicted in the Hokusai’s picture. It is with abstraction of the details, but then again, also with dynamic and ideal low sound. It lacks with the attack impact and the release decay of sampling drum reality, but the artists chose its ideality.

The direct motivation that made me decided to recreate the TR-808 is because it became possible for us to virtually provide a massive large-scale analog synthesizer like System-700 on a computer.

Another reason why I made this is because instead of an oscilloscope we now have powerful technological environment for analyzing a sound, which is Fourier transformation. As mentioned above, while we were developing the TR-808 we analyzed the sonic waveform by observing it through an oscilloscope. By the time we finished developing the TR-909, the microcomputers achieved 16bit processing, and in the lab we developed a tool for sound analysis known as FFT: Fast Fourier Transform. At that time, we were about to analyze the acoustic piano sound in order to develop digital pianos. According to the available reference documents then, there were suggestions of Short Term FFT by using 256, 512, and 1024 samples, but this doesn’t provide effective data that can be used as hints for synthesizing a practical instrument sound. Therefore, we developed TSPFFT or Time Spectrum Partial FFT (not short term FFT) that is capable of analyzing long samples such as those encompassing several seconds. This method utilizes the symmetry of the FFT time-space, and enables us to break down a sound into major partials that are dominant on both time and space domains. J. Kadoya who was in charge of this was able to analyze the entire piano sound with a limited environment that doesn’t allow floating decimal point computing by repeatedly delving down into lower digits of the fixed decimal point calculation. We were sorry to see it not making its way into TR-909 development, but it became the foundation for establishing the next era of digital piano SA synthesis method, and digital synthesizer LA method.  

An instrument sound is a sonic entity with time evolution, and hence cannot be analyzed with just auditory perception. For the sound designing of the RC-808, this analysis tool TSPFFT is recreated and is used for analyzing the acoustic sounds. The object in the Art of Painting is within the visual world of nature, and we first start from emulating them. The object in the Art of Sound is in the acoustic sounds again in the natural world, and the process of visualizing it and emulating it, is the creative activity to generate new stimulating sound. Hence it is the Art of Sound.