Super Audio CD (SACD) - is it just a load of noise?
SACD is a CD-like format that promises better sound for the consumer. Its professional equivalent is DSD, or 'direct-stream digital' recording. There are a number of systems that are able to record in the DSD format, which is very different to conventional PCM.
PCM (pulse-code modulation) is 'ordinary' digital recording used on CD and digital audio workstations. An analog audio signal is converted to PCM by sampling it several tens of thousands of times a second and measuring the voltage. The voltage is converted - quantized - to a digital value.
In the CD format, the quantization is 16-bit, the sampling rate is 44.1 kHz. For professional recording, we would choose 24-bit quantization with a sampling rate of 96 kHz. For the ultimate high fidelity, then 192 kHz sampling is also available in high-end recording equipment.
DSD turns this logic around completely. DSD only uses one bit quantization rather than sixteen or twenty-four bits. But it samples at a massive 2.8824 Megahertz (per channel)! This is way greater than even the highest sampling rate practical in PCM.
But the trouble is in the noise. Using just one bit leads to a signal-to-noise ratio of around 6 dB - even a wax cylinder recording is quieter than that. So clever 'noise shaping' is employed to make sure that the noise is placed above the audible band of frequencies, and what we hear is relatively noise-free.
But the figures just don't add up.
Take CD-resolution and multiply the 65,536 possibilities provided by sixteen bits, by the sampling rate of 44,100 samples per second. This gives around 2900 million possible data combinations per second. Now take DSD - multiply the 2 possibilities provided by one bit by the sampling rate of 2.8824 MHz, which gives around 5.8 million possible data combinations per second.
Plainly, DSD is not describing the signal anywhere near as precisely as PCM. In theory, DSD has an incredibly wide bandwidth. But the specification of DSD insists that the audio bandwidth be kept down to 22 kHz. That's not surprising - any signal above that is pretty much all noise!
There's another drawback. As things stand, DSD is incapable of being processed in any way. DSD uses a single bit, and digital EQ would create multiple bits, which would not be compatible. So to process a DSD signal, it has to be converted to PCM, then back again, which causes generation loss.
In conclusion DSD is a complete red herring. A total waste of everyone's time and energy. And SACD? In the words of the great John Watkinson it stands for 'Sad Alternative to a Compact Disc'!