You are here

Tape Noise Reduction

Exploration By Paul White
Published January 1996

Noise reduction comes in various guises, with Dolby alone offering five different systems. Paul White explains the difference between Dolby, dbx, and a bag on your head...

In the context of analogue tape recording, noise reduction is desirable because, without it, tape simply doesn't have an adequate dynamic range unless very high tape speeds and wide track formats are used. In practical terms, this means that the level of background tape hiss (which is a nominally constant level), is likely to become obtrusive during quieter passages where the level of the recording isn't high enough to mask the hiss. Noise reduction systems provide an effective means of increasing the usable dynamic range by reducing the subjective level of tape hiss during quiet passages.

The forerunner to modern noise reduction was a system known as pre‑emphasis/de‑emphasis, which works by applying a high frequency (HF) boost to signals during recording, then applying an equal degree of HF cut during playback. In theory, this restores the programme material to its original state while applying top cut to any tape hiss that may be present. Due to the response of the human ear, and because successive octaves of sound each double in frequency range (and hence the energy level of white noise also doubles every octave), random noise is most noticeable at high frequencies, which means that pre‑emphasis can bring about a worthwhile reduction in noise. The other reason that pre‑emphasis works as well as it does is that in most naturally occurring sounds, the high frequency signal components are much lower in level than the low frequency components, so a significant amount of high frequency boost can be applied to a 'typical' signal before the high frequency level comes up to that of the low frequencies. If this were not the case, pre‑emphasis would offer no advantages as any significant degree of HF boost would simply cause the signal to overload.

One thing that simple pre‑emphasis illustrates very well is that noise reduction comes in two stages: (i) the encoding stage, which is applied during recording; (ii) the decoding stage, where the opposite treatment is applied during playback. This is very important, because if the noise reduction system isn't used both during recording and playback, the tonal balance of the programme material will be drastically affected.

Having dispensed with introductions, let's look at the commonly used systems and see how they differ.

DBX

Most Tascam Portastudios and some Tascam open‑reel tape machines use the dbx system, which achieves very high levels of noise reduction by combining HF pre‑emphasis with 2:1 compression across the full audio spectrum; on replay, a 1:2 expansion is engaged along with de‑emphasis. Theoretically, dbx can achieve an impressive 30dB of noise reduction which means that even a cassette‑based multitracker can rival a 16‑bit digital recorder in terms of dynamic range. However, such a high degree of compression means that the replay signal must match the recorded signal very accurately, both in terms of level and frequency response, if decoding errors are to be avoided. Because of this restriction, two types of dbx system have evolved: Type 1 for professional use; Type 2 for domestic use, where some HF noise reduction is traded off against a greater tolerance of machine and tape accuracy. Type 2 is obviously the most suitable system for cassette machines, while Type 1 is better suited to open‑reel recorders.

Because dbx applies full‑band compression/ expansion, it is sometimes possible to hear hiss in the presence of low frequency sounds, where there is no high frequency content present to mask the hiss. For example, a mellow bass guitar played solo may be accompanied by a halo of noise which increases in level as the notes are played and decays as the notes decay. There is little that can be done about this other than to optimise your recording levels.

To get the best out of dbx, recordings should be made on the recommended tape types specified by the manufacturer, and the recording level should not be driven into the red but should instead peak at around 0VU. Overdriving the tape will simply increase the magnitude of any tracking errors, whereas with Dolby systems, pushing record levels into the red is almost routine.

Dolby B

The most commonly encountered noise reduction system is Dolby B, which is used extensively in domestic hi‑fi cassette machines. Dolby B has a lot in common with pre‑emphasis, but pre‑emphasis isn't applied equally to all signals — it only steps in when the programme level falls below a specific threshold. What's more, the frequency above which the boost is applied also varies depending on the spectral content of the programme material, though it never falls below 400Hz. Consequently, high level signals that are loud enough to mask the tape noise aren't treated at all, but vulnerable low level signals are subjected to HF boost. The maximum boost that Dolby B applies is 10dB, so it follows that the very best improvement that Dolby B can yield is a drop of 10dB in the level of hiss.

Dolby B is a level‑dependent system, so it is essential that the recorder is lined up to give the correct playback level, otherwise the decoding part of the system will be fooled into working incorrectly with the result that the reproduced sound will be tonally affected. Poor machine alignment is probably the main reason that Dolby B seems to dull the sound on some budget consumer machines, and it's almost certainly the reason that tapes made on one machine sometimes don't sound quite the same when played back on another, even though both may have Dolby B fitted. However, it's not only the machine alignment that can cause trouble — tape machines come set up for a specific brand and type of tape, and if you change types, it's likely that the replay level will be either higher or lower then the record level, which leads to similar problems. If you want to use a type of tape not recommended for your machine, you may need to have the machine specifically aligned for your choice of tape. This is true regardless of the type of noise reduction system you are using.

Dolby C

Dolby C is probably more relevant to home recording than Dolby B. Indeed, Fostex adopted Dolby C as their preferred noise reduction type for the majority of their cassette multitrackers and open‑reel machines.

The system works in a similar way to Dolby B but is effective right down to frequencies of 100Hz, and incorporates anti‑saturation circuitry to prevent HF overload when top boost is being applied to signals that already have a high level of HF content. If this was not done, the tape could easily saturate. This would not only add distortion but also compress the signal level, leading to inaccurate decoding.

Dolby S sounds more natural than either Dolby B or C and its noise performance means that even narrow format multitrack machines compare well with budget digital multitracks.

Dolby C does most of its work between 1kHz and 10kHz — the human ear is relatively insensitive to frequencies in excess of 10kHz, and by limiting the frequency range over which the system operates, mechanical alignment errors in the tape machine have less of a detrimental effect. In addition to being a generally more tolerant system than Dolby B, Dolby C provides up to 20dB of noise reduction. Since more noise reduction is available, however, the system is more likely to be compromised by the poor electrical setting up of tape machines or by the use of unsuitable types of tape. Note that with all types of noise reduction, the same type of noise reduction must be used for both recording and playback. Replaying a Dolby B encoded tape on a Dolby C system will sound tonally incorrect.

Dolby A

Dolby A has been around for the best part of 30 years and is only ever used in professional recording systems, never in consumer machines. Again the system works on selective pre‑emphasis, but this time the audio signal is split into four separate frequency bands (three shelving and one bandpass), each of which is processed independently and then added back to the original untreated signal. The main benefit of using multiple frequency bands is that high level sounds occupying only a narrow band within the audio spectrum don't compromise the amount of noise reduction being applied to the remaining bands. The maximum amount of available noise reduction is actually less than that offered by Dolby C, being about 15dB (only 10dB below 5kHz), but the multi‑band approach introduces fewer noticeable side‑effects than the simpler domestic systems.

Once again, decoding is the exact opposite of encoding, where the signals added to the original signal during recording are subtracted, leaving the original signal intact.

Dolby SR

SR is Dolby's flagship noise reduction system and was introduced a few years ago, around the time digital recording really took off, in an attempt to give the analogue recorder a longer lease of life. Up to 25dB of noise reduction is possible with minimal side‑effects, but Dolby SR is technically quite complicated and hence costly. Unlike previous Dolby systems which operated on the 'if it ain't broke, don't fix it' principle by treating only low level sounds, Dolby SR really goes for broke by trying to ensure that the maximum possible energy is recorded in all frequency bands all of the time. To do this, SR employs 10 filters, some with fixed frequency bands and others that slide to cover different parts of the spectrum according to the varying characteristics of the programme material. This means the system has to perform pretty complex analysis of the input signal in real time, and the decoding process is equally complex. Anti‑saturation measures are employed to prevent HF tape overload.

Dolby SR is heralded by many as enabling analogue recordings to sound cleaner and more transparent than the best digital recordings. As an additional benefit, it is reasonably tolerant of level errors or tape speed changes, which means that a degree of varispeed may be used without ruining the sound. Due to its high cost, Dolby SR is generally only found in professional systems, usually in the form of an add‑on rack system or as plug‑in cards.

Dolby S

Dolby S should really come directly after Dolby C, because it's a semi‑pro system, and like Dolby C it has found its way into consumer hi‑fi machines as well as some home recording multitrack recorders. It is related to Dolby SR, which is why SR was discussed first, but it would be wrong to consider Dolby S as being a low cost version of SR — in reality, it more closely resembles Dolby C, but with some of the filter technology from SR thrown in. Subjectively, Dolby S sounds more natural than either Dolby B or C and its noise performance means that even narrow format multitrack machines compare well with budget digital multitracks in terms of noise and sonic transparency. Though Dolby S is considerably cheaper than SR, it still costs more than Dolby B and C systems.

Summary

In the audio world you never get something for nothing, and all noise reduction systems have side‑effects that compromise the audio quality to some extent. At one end of the scale we have the costly Dolby SR system, which does a beautiful job of preserving audio integrity, while at the other end of the scale artifacts such as noise pumping, timbral changes or a perceived loss of 'life' are likely to occur to a greater or lesser extent with most consumer noise reduction systems, depending on the type of system used and how well the hardware is set up.

It must be understood that all the noise reduction systems discussed act only on tape noise — they have no effect on hiss that's already present in the material you are recording, and the same noise reduction system must be engaged both during recording and playback. It should also be understood that noise reduction systems only act to reduce the level of noise when the audio signal itself is at a low level. When the recorded signal is at or near its maximum level, the background hiss will not be reduced in level — we rely on the music to drown out or mask the hiss. Multi‑band systems such as Dolby A and Dolby SR are most effective, because they allow different parts of the audio spectrum to be processed independently; with simpler single band systems, narrow band audio in one part of the spectrum may fail to mask noise occurring in other parts of the spectrum.

Finally, noise reduction can only work effectively when the tape machine is properly aligned and when the correct type of tape is being used. This necessitates cleaning your tape heads before every recording session and having your machine serviced at sensible intervals. Used with care, even the simplest noise reduction system can bring about a worthwhile improvement in subjective recorded sound quality, so if the price you have to pay is a slight change in sound quality, think about the alternative — eternal hiss!