[LAU] 48k vs. 44.1k

Len Ovens len at ovenwerks.net
Mon Jan 7 04:37:11 UTC 2013

On Sun, January 6, 2013 5:55 pm, Ralph Bluecoat wrote:
>> Having decided that, what other issues cross my
>> path if I default my audio rate to 48k?
> Mainly, clocking issues. If for some reason an application thinks it is
> running in 44.1 but the system is at 48, there will be playback issues
> like
> pops and the likes.. I know I ran into this working with Reaper using
> Wine.

So wine was not told the correct clock speed. Funny, I ran into the same
problem... My laptop has a mic that is 48k only and it makes clicks and
pop when pulse is set to default at 44.1k.

> Perhaps not much of a 21st century issue, but if you are making projects
> that you intend to put onto CD, you will have to downsample to 44.1hz at
> 16-bit, as this is the Red Book specification for CD quality audio. A
> trivial task, but still time consuming and an extra step to think about.

Apparently not. Again that has been discussed in the last few weeks on
this list. It is just an export option during mixdown which has to be done
> Speaking as one of the devs for the Ubuntu Studio distro, I'm wondering
>> how much trouble I'm going to get for asking for 48k default sample rate
>> in audio.
> It really depends on what the target audience for the audio software. In
> the pro audio-music world, 44.1khz is still more widely used because it
> has
> been a standard for years in digital audio for CD, and I honestly don't
> see
> it changing to 48khz anytime soon (96/24 would be the next "level" you
> would typically find.) A 48khz sample rate is used (almost exclusively) in
> the film industry for synching audio to visual media rather than pure
> audio. Games, movie, and television audio are in a 48khz format at
> minimal.

The pro audio world as represented by this list seems to disagree at least
to some extent. 96k sounds nice and is used in live work like mixing and
such mostly because of it's lower latency. While things are changing there
are a lot of ADAT IFs that are 48/44.1k only, Those that handle 96k, do so
only with the loss of two channels. So 48k is very much still alive. The
semipro world is more likely to choose more inputs over 96k unless their
outboard gear can show the difference.

The internal sound IFs these days are designed around a bus/clocking
system that runs at 48k (actually a multiple there of) and making 44.1k
work requires lots of extra work. In the old ac97 series it seems inputs
and outputs are all 48K with a firmware layer in the onboard DSP for
pretending to work at other rates (read resampling). The newer intel HDA
systems will accept 44.1k raw, but with the bit clock they send out the
extra component cost to derive the right clock to do so is high
(expensive). There is no guarantee it is done with any quality. That is
not even dealing with the higher filtering costs (4x as high if I remember
right) already mentioned in the last two weeks.

Also, as happens, Ubuntu Studio does have a video production audience as
well. 48K would be correct for them too.

>>From a recording standpoint, it would be much more beneficial to change
>> the
> bitrate at 24 bits instead of the default 16, as you get more dynamic
> range
> and headroom in the recording process.

What comes from the card is what comes from the card. We can't change
that. We'll take 24bit by default if we get it already. However, from the
time it hits jack it is 32 bit (floating) and all 24 bits are preserved.
16 bit has not been the default for pro or even semipro (or consumer for
that matter) for 10 years or more. (I've had my D66, 24bit/44.1k to 96k
that long and it has never been "top of the pile")

Yup, CDs are 44.1k/16bit. But why record that way? 48k (or 96k if you can
afford it) is better than 44.1k on almost all computer IFs by design.

A week or two ago I would have said use 44.1k if that is what you want to
end up with. but there have been many convincing reasons for me to change
my default to 48K related on this mailing list (at least with the older HW
I have)... and as it doesn't seem to degrade normal desktop performance,
48K seems a good place to default to. Someone who has the gear is already
looking for how they have to change things to use 96K, but  default of 48K
should work with any non-broken SW. I have gone through all my audio
related SW on my system with no problems so far.

Len Ovens

More information about the Linux-audio-user mailing list