Additional technical information

39

 

 

source, the average and peak spectrum level declines with frequency above a few kHz. Best use is therefore made of the channel by pre- emphasising high-frequency signals to an extent where they are more likely to occupy the capacity of the channel. De-emphasis is applied on replay or reception and has the dual benefit of reducing noise and distortion from the preceding chain.

So far, all linear PCM standards used for digital audio permit the use of pre and de-emphasis; the universal characteristic being 50/15uS. Probably this standard came into play originally because it helped to alleviate the very real problems of linearity in early A/D and D/A converters.

More recently, the use of pre-emphasis has been neglected in digital audio. The main reasons for this being:

∙prejudice based on the very considerable difficulty of designing a transparent analogue pre-emphasis module in a microphone preamplifier.

∙the much-improved quality of A/D and D/A converters.

∙the availability of A/D conversion with more significant bits than the distribution channel (e.g. CD 16 bits).

This trend is mistaken, and for suitable material and conditions, a very real benefit can be obtained using pre-emphasis. The improvements with 518 are unique since the noise-shaping is designed for that circumstance and the pre-emphasis is effected in the DSP.

If we used pre-emphasis on the example of the previous section, the resulting CD could have a subjective dynamic range of 20 bits.

Simple rules for using 518

This section lays out some simple rules for using 518 in three different circumstances.

A step-down in sample word size

A step-down in sample word size while applying noise-shaping to give the channel a subjective dynamic range as close as possible to the input data. Such applications include transferring e.g. 20 bit masters to a 16 bit carrier like CD. In these cases:

∙Use Gain to bring peak level as close to full-scale as possible

∙If the programme material permits, use pre-emphasis

∙Use Shape B, C or D unless there is to be significant subsequent editing – in which case use either Shape A or high-pass dither.

No change in sample size

Modification of a signal in a channel where the same number of bits are used to define the input and output data. In this case we should be concerned with minimising the impact on the subjective dynamic range of the output signal. In these cases:

∙Use Gain to bring peak level as close to full-scale as possible

∙If the programme material permits, use pre-emphasis

∙Use Shape A or high-pass dither.

Step up in dynamic range

Shifting incoming data to allow the output to have potential for a subjective dynamic range that exceeds the input data without increasing

518 User Guide