Analog Audio Tape Performance & Possible Implications

What sort of performance can you expect from a first class, well maintained tape machine properly set up for the tape it is using? The most important parameters for characterizing an audio system are frequency response, distortion and signal to noise ratio. The more separation between channels, the better. With devices that have a mechanical component, various speed instabilities also matter.

If a system has a sufficiently flat frequency response over the 20Hz to 20kHz audio band (which is what the human ear can actually hear), a sufficiently high signal to noise ratio and sufficiently low distortion, it is likely to have no objectionable artefacts. It will (very probably) sound good!

But what is "sufficient"? For the delivery of music to consumers, the original CD standard is a good starting point. When this was introduced, experiments were done to determine what would constitute a "transparent" channel - something that added no audible defects under all reasonable listening conditions. The CD standard meets these requirements. At least in my opinion. (It has certainly come in for quite a beating over the last 30 years from some quarters, but I'll ignore that!). What performance do we expect from the CD format? Well, here is a typical player specification (a ReVox B225 from 1984):

  • Frequency response: 20 Hz - 20 kHz +0/-0.6dB
  • Distortion: < 0.006%
  • Signal to noise ratio: 96dB
These sorts of numbers come pretty much directly from the basic parameters of 16 bit samples and 44.1kHz sampling rate. In spite of endless arguments to the contrary, you can see why this specification is likely to be transparent for almost all practical circumstances. For example, a quiet room has a noise level of 30dB with respect to the standard 20uPa reference pressure. It is rare to find quieter rooms than that in domestic premises and most are noisier. If we match the electronic signal noise to the room background noise, then at 96dB above background, we will reach 126dB sound pressure level. The threshold of pain is 130dB and exposure to 125dB SPL for more than one minute is expected to cause permanent hearing damage. Personally, I'd say that 96dB signal to noise ratio is "adequate" for a distribution medium for end users! The other parameters are similarly "adequate".

How does analog tape compare? In a word, badly! Here are the sorts of numbers you can expect:
  • Frequency response: 20 Hz - 20 kHz +2/-2dB Sometimes, rather better than this.
  • Distortion: > 0.1% up to 1%-2% on peaks. Almost all 3rd harmonic though.
  • Signal to noise ratio: ~60dB, at best 70dB (depending on weighting filters). Maybe up to 10 dB better if Dolby (or similar) noise reduction is used.
Add to that:
  • Channel separation: 45-50dB maybe ...
  • Speed fluctuations: Wow & flutter >0.03% (weighted), at the very best. Approaching 0.1% is not unusual.
  • Tape modulation noise (a signal dependent defect specific to analog tape). This is usually unobtrusive, but that depends on the programme material.
  • Dropouts: The signal largely disappears if dirt causes the tape and playback head to separate. (This is actually not as common as you might fear on well maintained, high quality machines. Quite rare, in fact.)
And these sorts of numbers are for "professional", "studio" machines at 15 inches per second type speed. Things are not looking good.

Yet I claim that music recorded on machines of this kind very often sounds very good indeed. Of course, I may simply be wrong. However, pretty much everything recorded between the early 1950's and the early 1980's (at least) was recorded on such machines and a lot of that (now distributed digitally) can be easily obtained so you can judge for yourself. Some of this material has been "digitally remastered" and that may include some additional noise reduction but much of the benefit of remastering is said to be due to using the first or second generation analog master tapes. (Remastering can have a downside ... material may be processed to have reduced dynamic range and hence sound "louder". This is probably not a good thing, in general, even though it seems to boost sales).

You certainly can hear the effects of the relatively poor signal to noise ratio of tape ... tape hiss is quite audible. But, perhaps suprisingly, it is rarely intrusive. After a short while, you tend not to notice it unless you consciously listen out for it.

So, if the defects of analog tape are not generally objectionable ... what would be? I would guess the following:
  • Significant amounts of higher order harmonic distortion. The higher the order, the less is likely to be acceptable. Processes which generate a lot of high order harmonics (such as hard clipping) really do sound unbearable.
  • Significant amounts of intermodulation distortion which introduces frequencies which are not "musically (harmonically) related" to the original material.
  • Noise that isn't highly predictable (unlike the "white noise" of tape hiss). Pops, clicks and other joys of vinyl records spring to mind.
Now, you wouldn't deliberately design, say, an amplifier to match the performance of tape! But it does make you wonder when striving for perfection becomes an academic exercise given that some sorts of quite substantial defects appear to be quite acceptable to the ear. Approaching perfection might be fun! But improving, say, distortion from 0.001% to 0.0001% is hardly essential. It is very likely no one will ever hear the difference. (There is much evidence that low order harmonic distortion below 0.1% is inaudible). Perhaps if you are feeding the signal through a few hundred devices in series, it does become worthwhile.
So, given that analog tape sounds so nice, should the world give up on digital recording? Absolutely not! As a hobby, or as a very specific artistic choice, there is nothing wrong with using analog tape recorders. But there is nothing wrong (at all) with properly designed and implemented digital recording systems. Analog tape machines have (at least) the following disadvantages:
  • The objective defects already discussed. The overall sound may still be excellent (I claim it often is), but these defects are not optional with analog recording. You may rather live without them most of the time.
  • Generation loss. The first generation analog recording may sound great. Copying it to another analog tape will cause the inherent defects of analog recording to accumulate. This is something you really don't want.
  • The cost of tape. For quarter inch width you are currently looking at over 50 pence a minute at 7.5 inches per second! (Double that for "best quality" 15 ips). Compare that to the cost of hard disk storage ... which is approaching zero these days. Goodness only knows what 2 inch wide tape costs.
  • The availablility of tape. At the time of writing (mid 2012) there are only two manufacturers of analog tape, and one of those (RMGI) recently sold its operations to another company (Pyral, which is maintaining production). The other (ATR Magnetics) may be a very small scale operation. I wouldn't bet much money on either of them supplying tape in 10 years time.
  • Maintaining and servicing the machine. No one makes analog tape machines any more. Probably no one will ever make them again. For some machines there are a lot of spares still available, but, especially for the more exotic studio machines, this will be an ever increasing problem.
  • Setting up the machine for a type of tape and compensating for head wear. Actually, this isn't that big a problem. Given some test equipment and reference tapes it is quite straightforward, and good machines do not go out of alignment very quickly. Things do have to be kept scrupulously clean, but it is really not that hard.
A digital world is better than an all analog world. A bit of analog in a digital world may be nicer than an all digital world, though. More interesting, anyway.

And while we are on the subject of digital audio ... well, you have to wonder just what is and isn't audible under practical listening conditions there too. This page has some thoughts on that.
Note 1: Some references to experiments on the (lack of) audibility of defects in 44.1kHz 16 bit audio are:
  • "Audibility of a CD-Standard A/D/A Loop Inserted into High-resolution Audio Playback", Meyer, E. B., and Moran, D. R., J. Audio Eng. Soc. Vol 55, No 9, 2007.
  • "Which Bandwidth Is Necessary for Optimal Sound Transmission?", Plenge, G., Jabukowski, H., and Schoene, P., J. Audio Eng. Soc. Vol 28, pp 114-119, 1980.
  • "The Digital Challenge - More on ABX Testing", Lipshitz, S. P., Boston Audio Society, 1984.

Note 2: More complex digital processing for use with analog tape has been developed. A particularly clever example is the "Plangent process" that corrects for mechanical speed variations. See:
  • "Correction of Wow and Flutter Effects in Analog Tape Transfers", Howarth, J. & Wolfe, P. J, Presented at the 117th AES Convention, 2004.
I have no idea how often this process (and maybe there are other, similar ones) gets used. It certainly did not get used on the 2009 remastering of The Beatles (see Sound On Sound, October 2009). The original master tapes were played back on a Studer A80 and not very much processing was done on them at all (noise reduction was only applied to gaps and fades, for example).
Note 3: Actually, I may be wrong here. It is possible that the Otari MX-5050BIII is still a current product. See: Otari's web site for more information. It isn't clear where you would actually buy one from though.

Go home ...