Clearly real instruments have analogue outputs (especially the voice), yet nearly all music is stored and sold digitally. The recording process is ‘perfect’ after decades of development, the human ear is well understood – yet vinyl records are coming back.
Analogue recording requires multiple types of signal conversion – from pressure waves to electrical signals and finally to magnetic domains written by a tape head. Each stage of conversion will introduce a range of noise sources, from the familiar hiss of white noise to the rumbling of motor bearings. Controlling and minimising the influence of these noise sources has always been a key problem for sound engineers.
What's in the secret analogue sauce?
If analogue is noisy and digital is ‘perfect’, why do some people prefer vinyl? Assuming it’s not just nostalgia or that the effort of putting a record on makes one listen to music more carefully, it must be down to the sound quality. And since analogue engineers were very concerned about noise, something they did to reduce it ‘back in the day’ must have made the sound better than it really should be.
Although noise can be reduced by good design, there are practical limits. Some types of noise are generated by heat, and nothing can be done short of cooling the electronics. Other types (particularly pink or 1/f noise) are a limitation of the electronic devices used, and were particularly bad in early transistors. Although recording studios could buy high end hardware, the same physics and limitations applied and the signal to noise ratios achievable were still limited. At the playback end, the limits were even greater since consumers expected devices to be affordable and compact.
Understanding the soundscape
If the noise can’t be made smaller (and by assumption the recording can’t be made louder) then how did the engineers of the day squeeze the technology harder? The answer is that they went in pursuit of good sound rather than dry technical numbers. The classic stages of compression and equalization were designed to optimise the flow of information to the ear by subduing sounds which would make noise apparent. For instance, tape noise reduction worked by boosting high frequencies in quiet passages (increasing their SNR) and cutting it during loud passages. This exploited the characteristics both of the ear (less sensitive at high volume) and the magnetic tape’s limited dynamic range.
In fact – much of the sounds of the era were shaped by the technology. Vinyl recording had a tricky balance between being too noisy and having too much distortion to make, and overcame it with clever equalisation. Tube amplifiers had other technical constraints (feedback versus clipping) to overcome which gave the classic ‘warm sound’. These classic sounds aren’t coincidence; they are in fact highly engineered.
The same engineering skills are used in a wide range of systems – from radar systems to industrial sensors. The electronics may have much less analogue content and digitise the signal as soon as possible, but the signals need to be filtered and processed and given to the user in an efficient, competitive way. The biggest question for any design is ‘where is the split between analogue and digital?’, and the answer is always ‘optimise the system for what the customer hears’.