Subject: Re: [linux-audio-user] Re: producing a drum sample library for hydrogen
From: anahata (anahata_AT_treewind.co.uk)
Date: Sun Jun 20 2004 - 17:17:28 EEST
On Sun, Jun 20, 2004 at 01:11:00PM +0100, tim hall wrote:
> Most people aren't aware of much above
> 16k, however the ear/brain is surely capable of perceiving differences, so a
> higher sample rate is going to sound smoother in the way that faster film
> looks smoother, the ear will perceive curves rather than digital grainyness.
There's certainly some evidence in favour of that, but consider this
counter-argument:
A 'grainy' signal could be regarded as the sum of a perfect signal plus
a small distortion signal. If you can demonstrate that the
distortion signal is inaudible then arguably it also doesn't have an
audible effect whan added to a sound that is audible. In fact the well
known auditory phenomenon of masking shows the reverse: a
sound that by istelf *is* audible can be rendered inaudible in the
presence of a simultaneous louder sound.
This principle should work whether the distortion of grainines is due to
lack of bits, where the distortion component should be a too low a level to
be audible, or sampling rate, where the distortion component is at too
high a frequency to be audible by itself.
My personal view (not an original idea) is that many of the perceived
differences are due to comparitively gross differences elsewhere in the
audio chain, which have not properly been taken into account. For example:
- A system with lower SR may have some in-band aliasing products because
the filters weren't good enough and there was some spurious HF content
lurking in the original signal.
- Brand A 24/96k input stage may have better preamps and other analog
circuitry than brand B 16/44.1k.
- An audio power amplifier may be nonlinear at frequencies between
20-40kHz. This is out of band but energy at those frequencies could
produce in-band intermodulation products.
- I read an anecdote from audio researcher Peter Baxandall in which
'subtle sonic differences' between two systems were traced to a 1dB
difference in level alone - which goes to show how subjective this
all is:-)
> Myself, I work at 16/48k, because my system is limited, not because I think
> it's best. Considering that, up to 2 years ago I was still using cassette
> tape, the sound quality I'm experiencing now is a freakin' revelation!
Also considering much of the world is cheerfully listening to compressed
MP3, minidisc, digital radio and MPEG audio, the numbers who really can
tell the difference or care very much are quite small. I do agree,
however, that the pro audio studio needs sound quality overkill because
of the many stages of processing the audio signal goes through (the same
applies to analogue audio), and when you are trying to assess possible
problems with the sound you need to be confident that your equipment is
beyond reproach in terms of sound degradation.
-- Anahata anahata_AT_treewind.co.uk -+- http://www.treewind.co.uk Home: 01638 720444 Mob: 07976 263827
This archive was generated by hypermail 2b28 : Sun Jun 20 2004 - 17:08:26 EEST