0

I was wondering about the effect of audio drivers on mixing music.

When mixing you ideally use studio monitors because they are supposed to have a 'neutral sound' so that you can achieve a mix that sounds okay on various types of devices (good headphones, cheap laptop speakers, smartphones, etc.).

However I am unsure about the influence of the employed audio drivers on decisions made during mixing. Given that different manufacturers of audio interfaces have their own native drivers I suspect that it's possible that the same raw audio files will sound different on different drivers? How do you account for variation while mixing? For example, I notice quite a substantial difference when I play my music using my laptop's native MME/DirectX drivers versus ASIO4all driver's (which I use for recording because of low latency). Funny enough, the ASIO4all drivers sound worse than the MME/DirectX drivers... Also, suppose you are using a super expensive (> 1000 $) audio interface for mixing. Then it's fair to say that the drivers of that interface will be quite good. However, if you're making music that people will listen to on all kinds of crappy audio devices with crappy audio drivers, does it matter that you mixed it on a high-end audio interface as the crappy audio drivers of the end user will eventually have to render the sound?

I hope my question is somewhat clear. Thanks for your time!

Regards, Bart

1

Funny enough, the ASIO4all drivers sound worse than the MME/DirectX drivers

That can't be. Something is broken if this is the case, or else you interpret a colouring effect as good that actually should be interpreted as signal degradation.

ASIO drivers (like ALSA drivers on Linux) should give the DAW straight access to the interface's DACs. Thus the DAC produces exactly the sound the DAW intends to, which because we're purely in the digital realm here, can be said to be perfect reproduction.

The main flip side is that this relies on giving the DAW complete control over the parameters like buffer size and sample rate and that nobody else messes with the contents of those buffers while the DAW writes to them, i.e. the output will be blocked for other applications. By contrast, sound servers like MME/DirectSound/XAudio2, CoreAudio or PulseAudio enable multiple applications to use the same DAC at a time. They do this by resampling, re-chunking and mixing the streams, which is not really a lossless process. Audiophiles for this reason often say things like “never use PulseAudio, because resampling is evil”. Frankly I consider this a bit silly, because properly implemented Lanczos resampling is virtually lossless for all audible frequencies (provided neither source nor sink use less than 44.1 kHz). In fact the master of modern records will usually be resampled already anyway, e.g. mixed in 96 kHz but distributed in 44.1 kHz.

So on a properly working system, ASIO and XAudio2 should effectively give the same output, at least any difference would be negligible compared to the differences that room acoustics or different speakers make. The main advantage of ASIO is, as you say, that it allows lower latency, because there's no extra processing.

I could think of some reasons why you might nevertheless think that XAudio2 gives better output:

  • You have set the latency too low? Obviously, the processor load is not only due to a sound-servers resampling; if you're mixing a record it will probably be mainly due to effect plugins. And if the CPU is already quite loaded, it will not always be able to finish all intermediate buffers so quickly before the next DAC interrupt; that's then a buffer underflow. Normally, this results in a distinctive drop-out that indicates what the problem is, but I could imagine that Ableton, being optimised for live performance, does something to make it still work smoothly albeit with degraded fidelity.
    For mixing and mastering, it generally makes sense to set the latency to something high: 512, 1024 or 2048 samples. XAudio2 will use similarly large buffers behind your back anyway, so there the problem doesn't arise.
  • Your interface doesn't really support the sample rate you try to use it at? Maybe the manifacturers only tested it properly at 44.1 kHz, then if you use it at 48 kHz there is some ringing of a not properly configured filter. With the XAudio2 drivers this would not be an issue because the signal is resampled beforehand, and as I said this can actually be done with quite good precision. ASIO on the other hand would force the DAC to run at your selected rate.
  • Your mix is too quiet? The old problem: louder sounds better. With ASIO, you must make sure your mix master actually stays withing the 0dB limit, and that's then sent straight to the DAC. Whereas a sound server, because of the processing it does, will actually have some limiter built in, and extra loudness controls. So it's possible that when using XAudio2, your actually listening to the signal too loud, louder than you can render it into a file. This may then appear to sound better, but really it doesn't, and if you mix or master this way the end result will be worse.
  • Hey, thanks for your elaborate answer. I think you're right about the reasons that I first thought ASIO sounds worse than MME/DirectX. First of all, I checked and the MME/DirectX drivers are indeed louder. Furthermore, I noticed that they also color the sound quite a bit because apparently the standard settings of my DTS sound studio feature a 'bass boost' and a fairly bottom heavy equalizer (so the sound had more 'punch' which I interpreted as better). It's also a good suggestion to increase the buffer size for mixing purposes as latency doesn't matter at this stage anymore. – Bart Michiels Jul 15 '18 at 10:33
  • So just to be clear: for mixing purposes I should always use the ASIO4all drivers right? You said in your answer that ASIO drivers allow for 'perfect reproduction' of the recorded sound so that means they don't color the sound at all and are thus ideal for mixing? Going back to my original question of what effect audio drivers have on mixing is the answer then that: as long as you're using ASIO drivers they don't have an influence at all on the sound? Also: is there a difference between ASIO4all drivers and commercial ASIO-based drivers that come with (expensive) audio interfaces? Thanks! – Bart Michiels Jul 15 '18 at 10:39
  • You should use a proper audio interface (not necessarily a super expensive one, but it should have reliable characteristics, which a consumer sound card doesn't) with the manufacturer-supplied ASIO driver. For mixing, use some neutral-sounding monitors. For mastering, you should ideally switch between multiple different speaker configurations, including both bass-heavy hi-fi and small thin-sounding monitors; it should sound at least decently on all of those. Of course, a purpose-specific mastering studio allows the most reliable results. – leftaroundabout Jul 15 '18 at 10:45
  • Ok, thanks for your advice! Do you happen to have a good recommendation for a 'proper audio interface'? I'm currently using a Line 6 UX2 and although it serves my current I/O requirements, I'm not really to pleased with how it sounds... – Bart Michiels Jul 15 '18 at 11:54
  • I haven't used any of the Line 6 interfaces, but generally, the sound differences between the professional-level interfaces are very small, at least as far as the outputs are concerned. The main difference is the quality of the mic inputs and stability at low latencies. You can't really go wrong with something from Focusrite, PreSonus, M-Audio or Tascam. – leftaroundabout Jul 15 '18 at 12:25
0

The assumption that monitors are neutral isn't really correct. It is important to know the sound of you speakers and the setup you are mixing on and mixing accordingly.

The reason the end of the process is mastering is that a good mastering engineer will fix anything that is out of whack from the mixing process.

It is common, if you are mixing in an unfamiliar studio, to bring a set of songs you intimately familiar with so you can understand the how the studio sounds.

As far as expensive speakers while mixing, it is very common to test the mix on some cheaper speakers to make sure it sounds good everywhere. It is also common to listen in mono to make sure nothing is lost. The studio I worked at every had a low power radio station (with all the compression that radion stations use) so you go in the parking lot and listen on your car radio. It wasn't used any more since it is now common to just burn a cd and go to the car and listen.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.