Meridian Audio Speaker manual What is dither? Why do we need it?, Digital Link, DSP Dimension

The Meridian Papers - 1 Meridian Loudspeakers: The DSP Path page 7

What is dither?

Why do we need it?

In the analogue world, as a signal dies away, it does so smoothly. As the level drops, the signal gets progressively qui- eter. At some point it reaches the same level as the noise. But importantly, if the signal level continues to drop, you can still hear it, despite the fact that it is below the noise floor.

In the raw digital environment, every- thing is different. What happens when a bit changes between a one and a zero is essentially inaudible at high levels, because there’s so much going on. But if you’re dealing with low-level signals, such as reverberation dying away, or the fade at the end of a track, the transition of bits from zero to one and back again becomes increasingly important.

As the level of a signal drops, it is repre- sented by fewer and fewer binary digits, and the changing of these bits becomes increasingly noticeable – it’s called “quan- tization distortion”. Ultimately, you sim- ply run out of bits, and when this hap- pens, the signal just stops, and in a 16-bit system such as Compact Disc, this hap- pens at an audible level. This behaviour is another of the several factors that gave early digital recordings a bad name, and led some pundits to claim, erroneously, that digital audio was fundamentally infe- rior to analogue.

Quite early on in the history of digital audio, it was discovered that a solution to the problem was to add noise to the sig- nal. At low levels, the effective result of this procedure is to turn the last few bits on and off at random, smoothing out the sound and ensuring that everything will not simply disappear as the level falls.

This noise is referred to as “dither noise” or simply “dither”. Truly random (white) noise, called “flat dither”, contains all fre- quencies and is clearly audible. More com- monly, a noise spectrum where the ener- gy in the highs and lows is modified somewhat (“triangular dither”) is used to make the dither noise smoother and more

benign… ➤

Being able to deliver bass from an enclosure an eighth of the size is a use- ful ability, because it enables us to make loudspeakers that are physically the right size: on a human scale (see later). That means that they will fit into a room more easily, and take up less space.

But it also means that stereo and sur- round imaging will be significantly improved, because the nearer to a point source your loudspeakers are, the more they can be made to disappear when recreating a soundfield in your listening room, especially when using Meridian’s advanced decoding technology such as Ambisonics and Trifield.

The Digital Link

Once we began to see the introduction of widespread digital music distribution media, from the Compact Disc (1984) to the Digital Versatile Disk (DVD) and the latest DVD-Audio specification, there was another step we could take: we could keep signals in the digital domain for as long as possible: from source to speaker (see sidebar: Digital Audio: Music to the Ears).

When it comes to active loudspeakers, a digital design has a great deal to com- mend it. To begin with, there are no long analogue cables carrying line level signals to the loudspeakers, with the possibility of induced hum and noise. Instead, a slim cable carrying a single channel of digital audio data is all that’s required (actually, we add a second cable to carry communications signals between the different parts of the sys- tem, but it doesn’t carry audio).

Full level is supplied to the loudspeakers, removing potential problems with noise at low signal levels, and the loudspeaker now has a user interface, if only to con- trol and indicate the volume – in fact Meridian loudspeakers have a display that can indicate a number of system parameters… or be turned off.

The DSP Dimension

Now we come to what is – for now – the final step on the path, with the addition of digital signal processing (DSP) to the digital loudspeaker. In Meridian DSP speaker designs, digital signal processing is used to implement the crossover.

This means that the digital to analogue converter (DAC) can be placed even later in the chain. In fact a separate DAC is used for each band of the loud- speaker system, maximising the system’s dynamic range – a current system can deliver up to 120 dB – giving better intermodulation performance, and offering a level of background noise below 10 dB SPL.

This, however, is only the beginning. With digital signal processing on board, you can do a great deal more. For example, you can design “impossible” crossovers, with linear phase, steep slopes and time delay compensation.

And thanks to highly accurate phasing between the drivers, the “beam” of the system can be steered for the best expe- rience at the listening position: there is even an axis control that allows the mid- range frequencies to be precisely tai- lored to suit your height.

Volume control in a Meridian DSP loud- speaker is handled by a precision combi- nation of analogue and digital tech- niques, combining to give the best of both worlds.

In addition, “balance” is not simply a matter of changing the relative levels of the speakers. Instead, the balance con- trol is a “Where am I sitting?” control. If you are listening to a stereo system and move to the left, you hear more level from the left speaker because you are nearer from it, and further away from the right, from which you hear less. But that’s only part of the story.