Introduction
The level of an audio signal would seem to be the most basic of measurements, and yet widespread misunderstanding and disagreement about programme levels has become arguably the biggest single obstacle to high quality sound reproduction.
Live sound covers an enormous range of levels, but this is not something that can be demonstrated with a conventional sound level meter. Sound level meters respond quite slowly, even on a ‘fast’ setting: they have to do because they use an RMS (root mean square) rectifier, which by definition must take a slow running average of the square of the input voltage. Music is complex, and constantly varying, with brief peaks originating from the initial impact of sticks on cymbals and drums, and a loud band might measure 100dB SPL on a sound level meter, yet have peaks reaching 130dB SPL or higher. It is these peaks that a recording system must handle, and they can be measured using a peak responding meter with an integration time of 0.5ms or less (not a standard IEC type PPM which has a longer integration time). Of course there is no such thing as the loudness of an orchestra or band: it depends on distance. Here we assume a desirable listening position (not actually among the instruments where ear protectors are recommended).
The sound level meter is also useless for properly assessing noise levels, since the commonly used A-weighting is based on equal-loudness contours for pure tones, and is not valid for the random noise. The subjective loudness of noise is best measured using a noise-meter to the ITU-R 468 standard. The chart below shows, on this basis, the real range of live music, and then the level capabilities of various stages in the audio chain, from microphone to loudspeaker.
The Problem
This chart is based on the assumption that what goes in should come out; true high-fidelity, and so an ‘Alignment Level’ corresponding to 100dB SPL has been assumed throughout. Any lower level would imply severe clipping at the first stage; the master recording. With this assumption, it can be seen that top quality microphones do not present a problem; most will handle 130dB SPL without severe distortion, and some manage 140dB SPL.
The master recording process, using current 24-bit techniques, can be seen to offer around 99dB of ‘true dynamic range’ (based on the 468 noise measurement); which is identical to the dynamic range of a good studio microphone, though it should be noted that very few recordings will use just one microphone, and so the noise on most recordings is likely to be the sum of several microphones after mixing, and probably at least 6dB worse than shown. Allowing 24dB of headroom, the highest peaks will be clipped, indicating the desirability of using a soft-limiter at the microphone input to attenuate the very highest levels. There is little point trying to record the highest levels faithfully, since they will not be reproducible on any known currently available loudspeaker, even if they were passed through the rest of the chain! A modern 24-bit master recording aligned to 100dB SPL is therefore capable of handling amost anything, without any need for the recording level to be adjusted to suit the programme level, although some increase in level for quiet material might make the recording more robust (less dependant on the lowest noise levels being consistently maintained on playback).
After this, things get more difficult, with compact disc or other 16-bit formats requiring some degree of compression of the highest levels, to fit within the 18dB of headroom. This turns out to be irrelevant though, because a typical ‘hi-end’ loudspeaker is not capable of reproducing anything above 105dB SPL. Even very expensive professional monitors only manage 110 or 115dB SPL, and then only at 3m listening distance! Attempting to play a master recording through any known loudspeakers at original recorded SPL is therefore doomed to result in clipping of peaks and considerable distortion!
Partial Reprieve – Listening at Lower Levels
Part of the solution to this problem is to accept that most people, most of the time, do not want to listen to loud music reproduced at realistic levels, but are happy with a reduction of at least 10dB. Making alignment level correspond to 90dB SPL at the listening position thus makes possible reproduction with 15 to 20dB of headroom, and on the very best professional monitors the maximum level of a master recording will just about be reproduced without clipping. It is for this reason that headroom should be maintained throughout the chain as far as possible, with 18dB of headroom on the CD. However this approach brings a serious problem, because low power speakers and portables will clip severely on peaks unless played at very low volume – too low to satisfy most listeners, especially when they want to ‘party’.
Compression and the State of Hi-Fi
This has led to use of severe compression on just about all modern ‘pop’ recordings. By compressing the brief peaks, a ‘punchy’ sound is obtained that can be turned up loud on average systems without sounding severely distorted. This means that the lost peaks, and true dynamics can never be restored, so that even when played on the very best system all modern recordings lack the ‘sparkle’ that was contributed by fast peaks in the live sound. They sound, in fact, like early analog recordings in which soft limiting automatically occurred, even at the first stage of master recording, as an unavoidable consequence recording to the (then ubiquitous) analog tape.
CDs and ‘Digital Sound’
When CD’s arrived, some thought they were artificial; too bright and ‘electronic- sounding’; others liked the new brightness, but few realised that it was not just a super-flat frequency response, and ultra-low noise level that was making the difference. Studio engineers had been recording at lower levels for some time, now that Dolby had reduced the noise level of tape drastically, so fast peaks were being preserved more on master tapes. CD had an even lower noise level than the master tapes, so these could be transferred directly to the new medium without losing anything – provided that full-scale digital-level was reserved for the accommodation of fast-peaks only, with the average level kept well down. Most early CD’s – Pink Floyd, The Beatles, had a ‘typical’ level around 18dB below maximum, and although most originated as analogue tapes they are to this day some of the best test material available to the listening public. Headroom had arrived, but not for long. Maintaining headroom meant that turning up the volume caused brief clipping on all but the most powerful systems, causing a harsh sound that had not been noticed on analog recordings. Many blamed some mysterious property of the digital recording process, and poor convertors and lack of proper ‘dithering’ were indeed to blame for roughness and granularity at low levels for a while, but it is probable that the main reason for the change in sound was increased headroom combined with systems unable to cope with it.
Audiophiles brought back the valve (tube) amplifier, which from the earliest days had been known for its smooth ‘punchy’ sound when used as a guitar amp at high volume. Valve amps, like analog tape, tend to soft-limit, so the harsh clipping was being ironed out at the last stage. This is a perfectly good solution, except that instead of returning to unreliable, inefficient valves, it would be better to provide a simple soft-limiting option in every power amp, using simple low-level circuitry. Instead, we are now stuck with low-headroom productions, a lack of dynamics in popular music, and a lack of sparkle on loud sections.
The ‘Over Volume’ Problem
Traditionally it was possible to turn up the volume control on most radios and audio systems to a higher level than the speakers and amplifier could handle – they had around 12dB of ‘over volume’. This meant that quiet passages could be turned up to an acceptable level, even though a loud passage might necessitate reducing the volume for acceptable distortion.
When CD manufacturers started to apply severe compression, however, they saw no point in leaving the upper digits on the discs unused, so they turned up the level, even though this brought no benefit since the level of noise on the CD was now so far below the noise on the recording as to be completely irrelevant.
Unfortunately the manufacturers of players, finding that all the latest recordings now sounded loud – so loud that the volume control had to be handled very carefully – stopped providing ‘overvolume’ and started to adopt the principle that peak recorded level, at full volume, should just clip at the output or power amp. The result is that all early recordings; the good ones with real headroom, sound very quiet even at full volume on most modern players, and this problem will be found to exist on almost all MP3 players, audio systems, and PC’s. This is a disaster! Overvolume, and the ability to clip, is essential on any system not capable of handling realistic peak levels (that’s all of them below 50kW per channel!) though it would of course be better if the clipping was properly handled for minimum audibility in a soft limiter.