Do you change Audio Midi Setup settings based on song bitrate?

[sound_wave]

Hello everyone. Ever since I've started using my Mac as music source, I've been enjoying my music listening a lot more. One thing that bothers me though is that I have to switch the Audio Midi settings every time I switch between 24/96 songs and 16/44.

 

I'm curious to find out if fellow forum members actually change audio midi settings based on the song's bit rate or just leave it at 24/96?

 

[BobH]

on a PC, but it amounts to the same thing anyway. I now leave Quicktime set to 24/96. My sound card will go higher but, firstly, I don't possess anything higher than 24/96 and, secondly, I was not impressed with the sound when I went that high. Between SRC to 24/96 and no SRC I cannot reliably tell the difference, so I may as well save myself some hassle when it comes to playing the high res stuff.

 

If anything, I think things are a tad clearer with SRC invoked but, as I said, it's so close as to not matter anyway.

 

[sidssp]

Whatever sounds good to you will be fine. I myself run Pure Music in HOG mode which basically takes over control of the digital output to the DAC, bypassing the Audio Midi settings.

 

 

[wgscott]

I leave mine at 24-bit,/48kHz, since my Nova can only sample at 96kHz. (I don't get sound if I set it to 96kHz, which would require the nova to sample at a frequency of 192 kHz).

 

The vast majority of what I have is sampled at 44.1kHz and has a bit depth of 16, so the principle is the same. I don't understand why people need to toggle back and forth, unless they have multiple DACs with different sampling frequencies. I really don't think you can hurt anything by leaving it at the highest settings compatible with your DAC. (I've asked the same question here, and have gotten that answer).

 

Human hearing only goes up to about 20kHz, and most people over 18 can't hear that high a frequency. So anything sampled at 44kHz should faithfully reproduce an audible sound wave (Shannon/Nyqvist theory). So I still can't figure out how people can claim to hear the difference between 48kHz and 96kHz, let alone higher sampling frequencies. So I guess the real question is really whether any harm can come from leaving the setting at (say) 44.1 kHz. I set it as high as it will go, just because I figure I have nothing to lose, but it is probably overkill.

 

[ted_b]

frequencies" it's because we have a DAC that is capable of playing any sampling frequency natively (i.e no up or down sampling). If you have a 24/96 file to play, then you want, in a perfect world, to hear that file processed at 24/96. If just using iTunes (i.e not using a music player like PM or Amarra which handles the Audio Midi switching automagically) then you would want to change the Audio Midi setting to 24/96 so your DAC could process it cleanly and natively. Many of us have DACS that play any bit rate and resolution depending on the signal coming in, and if that signal is not equal to its real value (i.e up or down sampled by iTunes, etc.) it's often a sonically detrimental compromise.

 

Sampling resolution is not the same thing as upper end frequency.

 

[wgscott]

So, why not just leave it at 96 kHz then?

 

Sampling resolution is not the same thing as upper end frequency..

 

Any signal (sound, light, etc) is a superposition (a Fourier sum) of sine and cosine waves, each of which has a unique frequency. Human hearing range is between 20Hz and 20kHz. Any components having a lower or higher frequency than this can be excluded from the superposition (summation) without changing what you hear. If the highest-frequency component you can hear is 20kHz, you can reconstruct that component faithfully with 40 kHz sampling, according to the Shannon/Nyqvist sampling theorem.

 

So sampling resolution needs only to be twice the upper end frequency that you hear. Anything above that cannot improve the signal in the audible range. I doubt it hurts to include it, but I can't see how it can help.

 

In any case, why can't you just set the Audio Midi to 96kHz and be done with it?

 

[BobH]

In any case, why can't you just set the Audio Midi to 96kHz and be done with it?

 

You can. It is what I do. The point in question is whether or not you want to carry out any SRC. Ted, along with many other folks around here, would prefer it if no SRC was taking place. In this case you need to leave well alone and adjust the Audio Midi to match the file being played. If, like me, you find the on-the-fly SRC done by iTunes to be acceptable then you can set it and forget it. Either way.

 

As for the limitations of human hearing, respectfully, they are not really an issue here. High sample rate is not so much about what we cannot hear, but the increased resolution of what we can hear. And the easier implementation of the final conversion to analogue - less steep filters and all that malarky.

 

There is no 'right' way, the only thing of import is that folks understand the implications of the settings they make. Setting Audio Midi to a resolution that does not match the one being played, will result in iTunes SRC'ing the output. As long as the user is aware of this, then fine. In other words, don't break the rules until you know what they are!

 

[The Computer Audiophile]

I always use the same sample rate as the music I play. However I see nothing wrong with setting it at a specific rate to make listening to music more convenient. I'll suffer the inconvenience.

 

[wgscott]

don't break the rules until you know what they are!

 

OK,

 

High sample rate is not so much about what we cannot hear, but the increased resolution of what we can hear.

 

http://en.wikipedia.org/wiki/Nyquist–Shannon_sampling_theorem

 

 

 

 

[ted_b]

Or downsample our 24/176. I thought, though, that we already said that? We want to listen to each sample rate natively....natively...no up or down sampling! For example, 16/44 to 24/96 resampling is tricky with less-than-stellar tools (like Isotope) cuz it's not a good integer multiple, for one. And downsampling great 24/192 to 24/96 is a huge waste of good resolution, let alone a now-processed signal. And we have PM or Amarra that does it for us, so no big deal.

 

[ted_b]

the greater musical quality of HiRez music files. It will save you lots of dough (DACS, storage, etc). Those of us who have heard the improvements are smitten, and poorer financially. Stay 16/44, leave your Audio Midi alone, and use whatever sounds good to you.

 

[wgscott]

24-bit greatly expands the dynamic range, so no one that I am aware contests whether this is an improvement (although you will only notice the improvements in instances where the original recording exceeds the dynamic range of a 16-bit recording). It just gives vastly more space, so there are no issues with respect to up-sampling.

 

As for whether you can hear the difference between audio sampled at 44kHz vs. 96 kHz (or higher), human physiology and the laws of physics are completely unambiguous. Maybe if I were a dolphin or a bat ....

 

[The Computer Audiophile]

Hi wgscott - You certainly are 100% correct about the limits of human hearing. However the limits of our hearing don't have that much to do with differences in sound quality between 44.1 and 96 kHz. It's my understanding that higher sample rates allow digital filters to be pushed up into the inaudible ranges along with a host of other engineering techniques. I am definitely not an expert in this area, but I've talked to many audio engineers about the subject. I only wish I could remember what they have all said.

 

Also, some DACs perform differently at different sample rates. For example one DAC I recently listened to extensively performed best at 24/88.2. This was according to the DAC designer and audible via listening sessions.

 

[almaatakz]

I keep the settins at 24/96. As I mainly use playlists and my library is a mix of hi-bitrate aac/mp3 (30%?) and alac of 16/44 (60%?) and 24/96 (10%?) manual switching is not practical. It really should be automated but itunes does not offer that and amarra mini is a bit expensive. I hope apple will update their os/itunes sooner or later to serve good playback better.

 

[Audio_ELF]

Apple have repeatedly stated that it is a decision they took NOT to provide automatic sample rate selection. They feel that there should be one selection for sample-rate (the Audio-Midi Panel) which is fixed. This is for "consumer" grade applications not recording / mastering applications by the way.

 

Eloise

 

[wgscott]

Thanks. That certainly makes more sense (as does the non-integer multiple for the upscaling ratio problem).

 

I've got a friend where I work who used to be an Apple VP and knows a lot about the internals. I'll see if I can pump him for some info...

 

[wgscott]

I'm having problems finding ANY documentation, even digging into CoreAudio. Can you point me in the right direction?

 

I can kludge together a "fix" but I would be shocked if there isn't a better way.

 

[Audio_ELF]

The big problem with the way iTunes, etc works (and I think this Apple have said is a bug rather than by design), is that iTunes reads the sample rate in Audio Midi when it starts (call this SR1). Any audio is then resampled (by iTunes) to this rate.

 

Unfortunately if you change the sample rate in Audio Midi (lets call this SR2), iTunes continues to resample everything to SR1 and then Core Audio resamples it to SR2.

 

So if you start iTunes with Audio Midi set to 44.1 and play a 96k track, iTunes resamples it. If though you change Audio Midi to 96k and play the same track in iTunes (without restarting iTunes), iTunes down-samples to 44.1k then Core Audio up-samples to 96k. If you have a display on your DAC, you think you're getting the proper 96k file- but instead it's been resampled twice.

 

Eloise

 

P.S. This is an old problem (back in version 7.x I first head of it) but I don't think it's been corrected since.

 

 

[wgscott]

I had NO idea. Version 7.X is like 15 years ago. The OS is completely different now (unix, totally different kernel, etc), so this is REALLY bad if the problem has persisted that long.

 

If a 96kHz track is sampled at 48kHz, presumably the sound isn't degraded (just sampled every other slice), but if a 96kHz track is sampled at 44.1 kHz, this sounds much more problematic.

 

[DanRubin]

I think he meant iTunes 7.x, not Mac OS 7.x.

 

I worked on OS 7 when I was at Apple way back when. It needed to run in 2MB of RAM -- imagine that!

 

[Audio_ELF]

@DanRubin ... yes I meant iTunes 7.x not Mac System 7.x Oh and HE is a SHE

 

@wgscott ... yes, sampling a 96k file at 48k isn't too bad, but sampling it at 44.1 can cause real trouble. In testing, the SRC that Apple utilises in iTunes is actually quite good (for on the fly SRC) but many people would argue against any form of SRC.

 

Generally I think a lot of people are waiting for a complete re-write of iTunes as it's (a) getting rather bloated, (b) isn't written in Cocoa and is only 32bit and © there are many slightly curious user interface decisions made in iTunes.

 

Eloise

 

[Miska]

For a proper SRC there shouldn't be any quality difference between 96 -> 48 or 44.1 other than the difference in sampling rate between 44.1 and 48.

 

Sure, the converter is not as easy to make and there are bad ones out there, but there doesn't have to be any quality degradation due to the conversion ratio. It's all just math...

 

Starting from 10.5, OS X offers reasonably good SRC. Same goes for Windows, Vista onwards.

 

Most home theather amplifiers do similar conversion internally, as well as various DACs. (upwards conversion in CA DacMagic, MuFi V-DAC, etc)

 

[Bob Stern]

Eloise said; "Apple have repeatedly stated that it is a decision they took NOT to provide automatic sample rate selection. They feel that there should be one selection for sample-rate (the Audio-Midi Panel) which is fixed. This is for "consumer" grade applications not recording / mastering applications by the way."

 

Actually, it's not a question of consumer v. pro. An Apple engineer on the Core Audio listserv said that sample rate switching should be performed manually by the user instead of automatically by an application because: 1. one application could cause another app to crash by switching the sample rate; or 2. two app's could disagree on the desired sample rate and keep trying to switch it back and forth.

 

 

wgscott said: "I'm having problems finding ANY documentation, even digging into CoreAudio. Can you point me in the right direction? I can kludge together a "fix" but I would be shocked if there isn't a better way."

 

The CoreAudio framework include a header AudioHardware.h that includes properties kAudioDevicePropertyNominalSampleRate and kAudioDevicePropertyAvailableNominalSampleRates.

 

[Miska]

An Apple engineer on the Core Audio listserv said that sample rate switching should be performed manually by the user instead of automatically by an application because: 1. one application could cause another app to crash by switching the sample rate; or 2. two app's could disagree on the desired sample rate and keep trying to switch it back and forth.

 

This is not a problem if one application has exclusive direct access to the device, this is what ASIO and WASAPI Exclusive mode specifically provide on Windows side.

 

In any case, especially for pro use, only one application should be able to access the audio device. It would be quite embarrassing to have an email "bling" sound in a live event being mixed into the live sound...

 

For audio processing pipelines, there are better alternatives.

 

[Bob Stern]

especially for pro use, only one application should be able to access the audio device. It would be quite embarrassing to have an email "bling" sound in a live event being mixed into the live sound.

 

On the Mac that's not a problem. In either the Sound preference pane or in Audio MIDI Setup, you can set the "default sound output device" to be your USB or Firewire audio interface or DAC while leaving the "system" output device set to the Mac's internal audio. The former receives audio from iTunes. The latter receives system alerts such as email "bing" sounds.