Digitising cassettes: Is 16 bit / 44.1 kHz enough ?
- Thread starter serendipitydawg
- Start date
Although the human hearing can go to 15kHz, instruments can produce harmonics up to 50kHz and beyond. This is what makes instruments sound the way they do in real life and one of the reasons digital has always been criticised due to the brick wall filtering at 22kHz.
Werner
pfm Member
Nope.
If one cannot hear above 15kHz, then one cannot hear the harmonics that reside above 15kHz.
The entire notion of harmonics is a gross simplification anyway.
Digital has generally been criticised by people that don't understand it, and to be fair, there aren't many people who do, as it is hard, and in the early days of commercialisation, the major players were keen to protect their IP. Bad visual representations of what is going on to try and explain the intricacies to the layman were interpreted as what was actually going on by the audio community, and that led to all sorts of nonsense being said. You'll commonly still see some sort of stepped diagram representing digital approximations of an analog signal with 'staircase' type output - it was ok in 1985 when trying to explain what was going on, but it is totally misleading.
Brick wall filtering is a similar issue - it doesn't occur in modern designs, and hasn't since the early 90s. You instead use sigma delta, noise shaping and decimation to avoid aliases. You could say decimation achieves this effect, but I don't think the criticism aimed at early analog domain high pole filters applies to the decimation filters of today (and that's on the ADC side anyhow).
Werner
pfm Member
And yet a certain John Atkinson in Stereophile used to explain it very well and very correctly in CD's early days. But once certain myths about digital audio arose he dropped his attempts to educate. It was clear what the audience wanted.
It most certainly features in modern designs, and for good reason, as it is prescribed in detail by the sampling theorem.
That is an entirely different thing. The early high-order analogue filters were a crude and failing attempt at approximating the required brickwall filtering. It is very funny that the 'apodising' and minimum phase digital filters that were so fashionable a couple of years ago (or still?) are nothing else than direct translations of the maligned early analogue filters.
John Phillips
pfm Member
For digitizing cassettes I am fairly sure there's no reason to go beyond 16-bits at 44.1 kHz because I doubt they capture anything more demanding. However, there is an interesting meta-analysis of multiple experiments on whether humans can perceive a difference between high-resolution audio and standard resolution audio. It is freely downloadable. I have scanned it but not read it carefully enough to assess its quality.
A Meta-Analysis of High Resolution Audio Perceptual Evaluation
On a quick read, the paper itself does not address what issues underlie what it discovers and whether it's high resolution per se or something that comes with high resolution.
Of course, it remains up to an individual to either chase his/her idea of perfection in audio perception or to declare "good enough" when the difference does not matter to them.
A Meta-Analysis of High Resolution Audio Perceptual Evaluation
"... Results showed a small but statistically significant ability of test subjects to discriminate high resolution content, and this effect increased dramatically when test subjects received extensive training. ..."
On a quick read, the paper itself does not address what issues underlie what it discovers and whether it's high resolution per se or something that comes with high resolution.
Of course, it remains up to an individual to either chase his/her idea of perfection in audio perception or to declare "good enough" when the difference does not matter to them.
That was the point I was trying to make, that a high order analog low pass filter (which is tricky to get right with limited passband ripple and a narrow transition band without expensive components) is no longer part of the DAC design, and this is what people in the audio community refer to as a 'brick wall filter'. I'm very aware that the output from a modern DAC is still bandlimited in exactly the same way, but it achieves it a rather different, and to my mind, more elegant way.
And yes, it's fun that the response of the filter implementation could be seen as equivalent to the analog filters that were previously panned. However, the joy of the digital implementation is that it behaves exactly as designed, year after year, without needing expensive components that will drift with temperature or change with age, so it's a much better place to be.