Panasonic SV-255 Portable DAT Recorder
This little DAT recorder was 48 khz, 16 bit. What I found appealing about it was the sound quality of it. It’s A/D convertors used a 1 bit delta sigma technology. This 1 bit information arrives at a very high sample rate, and then is converted into the PCM stream that is recorded onto the DAT tape.
One project I used with this recorder was a train ambience CD, called “The Course”. It sounded awesome! As a side note about that project, in order to catch the high dynamic range of passing trains, I would manually lower the input gain on the SV255 as the trains approached. When done correctly, this gave me a much greater signal to noise ratio so I could use the distant train nature ambiences when the train was really soft as well as the subsequent sonically brutal close pass-bys. This project was edited and mixed on an AMS Neve AudioFile/Logic system, which at the time was far sonically superior to the Digidesign systems in production then, and dare I say even the current Avid systems employed today.
A similar (identical?) front end one bit technology is used in DSD recording. http://en.wikipedia.org/wiki/Direct_Stream_Digital.
However in DSD there is no conversion into a linear PCM format. It remains at that extremely high sample rate. BTW for a sonic treat check out The Silk Road Ensemble with Yo Yo Ma. Some, if not all recordings were done in DSD. Beautiful.
Conclusion not quite yet!
I was going to end this article here, but I was thinking last night if there was any commonality to the three pieces of hardware that I had remembered from the past. What really was it that I was hearing that made a long lasting sonic impression on me?
Some years ago Digidesign, I think before it was acquired by Avid, ran a test online, an audio test. They had had four pieces of music produced in four different genres, rock, jazz, something else, and something else (sorry, forgetful Gramps talking here). Four different engineers each mixed one of the pieces of music, but they did it in two different ways. One was “in the box”, in other words completely digitally mixed, and the other was mixed through an SSL console. Both versions used multitrack recordings that were played back from a ProTools session, so the only way they differed was how these unprocessed digital recordings were combined into a two track master.
For the sessions that were done “in the box”, only an SSL strip plugin was used for EQ and compression, the same strip that had been emulated from the SSL console. The console mixed version only used the analog strips on the console, with no digital plugins. I don’t know whether they started their mixes with the console version, or the “in the box” version but once they had one finished, they took each track of the mix and set up the other version to be as close as possible in settings and levels.
Each track from the multitrack session was phase reversed and compared to the other version, fine tuned so that each was auditorially “zeroed out”. This assured that the setting in one realm would have the same sonic effect as the setting in the other realm.
The sessions then had the phase reversal removed as needed, were mixed as 48 khz 24 bit Wave files. The mixed files were then presented online so that one could download them without knowing which version was which, and then one could listen to them on their own system, and select online as to which version was the preferred, subjectively better sounding version. Once you submitted your results, you were then told which version you had selected.
Of the four mixes (each with two versions) that I listened to, I chose the SSL console version four times out of four. Cool, but a bit alarming to a digital nerd like me! I noticed a subtle but apparent difference in each mix, especially in the stereo depth. Also noticeable was the placement and clarity of any cymbal or other complex high frequency content.
So how does this Digidesign anecdote tie into the three older pieces of equipment I glowed about earlier? The commonality: high sample rates. In the Digidesign story, I think the SSL strip plugin is probably a really good emulation. Look at how the audio engineers were able to zero out the audio for each track. But the difference was in the mixing, the summing. An analog mixer has essentially a limitless sample rate. When signals are combined in a digital system, things have to fall into boundaries, the external sample boundaries, or the internal possibly higher rate boundaries. This can cause timing issues in the audio combinations.
Our ears are very sensitive to timing issues in stereo, and I think this also applies to mono things as well. Did dynamic range also play a part in the form of summing errors? Perhaps. In the three pieces of equipment I talked about before, each has a very high sample rate component inherent in it’s design. Each was capable of playing and/or recording something with an extended timing accuracy over conventionally timed digital devices.
Have current digital systems overlooked these timing issues in favour of power and convenience? If so, how high should sample rates go? I really don’t know. I can tell you I operate my recording sessions as reasonably high as I can go, which is at 96 khz, 24 bit, two times as high as 48 khz, but doesn’t begin to compete with the extreme high rates of the gear I mentioned earlier at 64 times, 256 times, and analog limitless sample rates.
Timing is Everything
The first standalone semi-modular from Pittsburgh Modular