“In accordance with the sampling rate of the 44.1kHz AD converter, we had to create filters that could surely eliminate sounds above 20kHz. Without that, it was possible to have noises when signals above the Nyquist frequency of 22.05 kHz were recorded. Also, a lot of higher harmonics had occurred after the DA conversion that we needed to filter out.”

“If steep filters were used to eliminate these higher frequencies all at once, a lot of distortion could happen. Our idea was to combine 4 second order filters and a single order first filter, combining for a total of 9 stages, necessary to filter out above 20kHz. At the testing stage, we had opinions such as “When high register instruments are playing above a certain range, it tended to become a sweeter tone, or sometimes could be heard as if the volume was getting smaller.” When we tried to figure out what was happening, we found that among the distortions occurring within the filters, higher harmonics above 20 kHz were eliminated and therefore, unheard. Depending on the input source, a large energy loss was happening. That had resulted in a reduction of the output level even though their input level was raised, and unfortunately, the sound got dull too. In order to overcome this problem, we had changed the types of operational amps used and their power allocations in order to avoid this “insufficient reproducibility of the recorded sound on certain instruments”. Probably, if we were testing only with sine waves, we couldn't have responded to what the recording engineers had commented on.”

“Talking about sound quality with digital recording, the S/N ratio had dramatically advanced from what we experienced with analog multi-tracks. We had to be careful on overloading the input level, but even with the low recording level, we never had a S/N ratio be dramatically worse. When we played back, no hissing noise could be heard and since the sound emerged from a very silent state, everybody was surprised the first time they heard that (laugh). Analog tapes have their own beauty in the unity of sound and its saturation, but when we consider the accuracy of the reproduction of the original sound, I knew the PCM-3324 was ahead. Even though it was digital and the recording was done on tape, we could not eliminate the possibility of physical damage to the tape, but those risks had been avoided by the error correction function.”

“For example, those instruments like drums, which we desired to have tape compressions added, were recorded by analog. But when it came to auto punch in/out, fast forward/rewind, and preciseness of the locator, the PCM-3324 was obviously ‘effortless’. Also, in times of analog tape technology, the recorded tapes sometimes transferred or lost some of their information while they were being stored and preserved. Of course, with digital recording, those problems were solved, and the sound was kept fresh until they were mixed.”

“Between the tracks on the tape, there were blank spaces called guard bands which suppressed the interference between the tracks due to magnetic flux leakage during the recording process. But when it came to incorporating 48 tracks, the width of the guard band had to become much narrower. As a result, the error rate became worse and we assumed that it would no longer be able to record. In order to solve this issue, we developed a system called “pulse train recording”. It was basically designed to not let the current flow into the adjacent tracks at the same time, quickly switching the timing of the current flow between adjacent tracks to avoid this interference between tracks. In order to reduce the magnetic flux leakage between the tracks, we had also developed a thin film recording head by applying semiconductor technology needed to replace conventional ferrite core designs.”

“Even as with the recording head, and that it was so hard to lower the magnetic interference between tracks, development of the playback head was an even more difficult task. The magnetic flux leakage was so horrible that it was almost impossible to extract the signals. So, we had a challenge to design the head as small as possible to suppress this magnetic flux leakage. But when we had gone to the factory at Sendai with our blueprints (now called Sony Storage Media Manufacturing), we were told that it was “too small to produce.”. But I kept asking, and that resulted in me joining the prototype production team, and we were successful in producing a head which could handle all 52 tracks, including audio and time code tracks. We then brought it back to Tokyo, and after installing it in the prototype of the PCM-3348 , the result was quite good. This was the morning of the day of the press announcement, so we were down to the very last minute (laugh).”

“I am proud to say that the PCM-3348’s sound quality was very good. With the PCM-3324, Ebata had created the masterpiece of analog filters, but their size was so big that when it came to 48 tracks, it wasn’t possible to be installed into such a chassis. In the mid-1980s though, digital filters were available and it turned out that we could get great sound, even with their compact size, if oversampling was done. We then incorporated that new technology. In this regard, the PCM-3324 and PCM-3348 therefore have very different sound qualities.”

“When this project started, in order to learn more exactly how recording studios operated, I had stayed several nights in the Roppongi Studio. What the engineers wanted us to learn was how important it was for us to provide recording devices that prevented engineers from making critical mistakes in their operation, even into the very late night hours that many recording sessions extend into. For example, in discussions about the layout of the knobs and switches, we were like “if we lay them out in this way, that could help in preventing any operational mistakes”, and we did various other similar refinements, as well. One thing I personally strongly pursued was the feeling of the touch given when the switches are being pushed. It was something that when the switches were engaged, we could say “Here you go!” to the recorder, and have complete trust that they were engaged. We invested some time and a bit of money to gather various switches from around the world, and then tried out which were the best. Also, since the PCM-3324 was already used in several places, we were thinking about compatibility between these two machines, so that the 24 track recorded tapes could be operated on the PCM-3348. As a result of this kind of design detail, at the AES show in 1988, Stevie Wonder actually touched the device and had praised it with the words “It's good!”. After its release, initially only planning to produce 400 units of the 3348, we sold out of them instantly. Since that was the time when the CD market was still growing, I believe that the PCM-3348 contributed in some part to making and capturing musical history.”

“It was very useable, and I’d even say that the layout of its operating controls were very appropriate. For example, when two switches are pushed simultaneously, it won't be useful if your arms were to cross. The appropriate layout was needed, and Sony’s engineers had learned so much from their time spent at recording studios, about the layout, how the switches are stroked, and the timings of punch in/out. The design engineers were clearly pursuing the best form.”

“Selected portions of the recorded sound were able to be sampled in the memory. For example, if the take of the first chorus sounded good and you wanted to use it on the second chorus, the sampling started when you engaged the memory switch while the intended take is being played back, and ends when you stop it at the intended point. When you trigger the sample at the intended point while playing the second chorus, you could record it on the chosen track. This is what we now call copy and paste. But at the time, if you were trying to do the same thing, you originally had to record on a different recorder, and then get it back to the digital multi-track. Also, the variable crossfade feature during punch in was very useful. For example, in slow songs when you wanted to punch in during a phrase with long notes, for example, you could smooth out the curve and modify it according to the source.”

“This was also available on the PCM-3324, but being able to use a full 48 tracks made a huge impact. This was because with an analog multi-track recorder, in order to do the mix using fader automation on the console, we had to record the time code on an audio track. As well, in order to avoid the time code interfering with the audio signals, adjacent tracks to the time code track had to be kept blank, so that it was really only 22 or 23 tracks, at most, that were able to be used. So, it was superb to be able to use all 48 tracks! Basically, the number of tracks were doubled from the PCM-3324, could use the same tape, and also a 24 track recorded tape was able to have another 24 tracks overdubbed, which made us really happy. It was very easy to add new sounds to older songs which were recorded with the PCM-3324, and we could also replace certain parts, too.”

“There was nothing unsatisfactory about the sound quality and it was a well-made product. After the mid-1980s, CDs had become the dominant music format and Sony products such as the PCM-3348 for recording, and the PCM-1630 processor or PCM-9000 disc recorder for mastering, were used all over the world. It made me proud to see the PCM-3348, in which I was involved in the development of, as one of the recording engineers, be so loved and widely used.”

“We eventually replaced all of our digital tape recorders with 11 Sony 3348HRs. This tape recorder to this day is the greatest sounding digital tape recorder ever manufactured, in my opinion. It actually would rival AND beat Pro Tools sonically in a side-by-side shootout.”