In what way is WASAPI exclusive to Audacity..?

[Micke-T]

Hi,

I’ve been doing LOTS of researching and reading and also several tests in order to find the “ultimate” method and settings for ripping vinyl and cassettes. Now the result basically sounds fine, but in my mind, there is a slight tendency towards a “harsh”, or “digital”, sound. So, I’m trying everything I can think of to make it sound totally transparent and now I’ve been exploring Exclusive mode for WASAPI and I have a question regarding the “exclusivity”.

I have a Focusrite Scarlett 2i2 (3rd Gen) as an A/D converter connected via USB to my ASUS UX430 Laptop running Windows 10. I’m running Audacity 3.3.3 and WASAPI and the Scarlett is set up (in Windows “Sound Settings”) to run in exclusive mode (see screen shots) and the quality is set to 24-bit 96kHz.

As can be seen in the screen shots, all other sound devices are inactive (Sorry about the Swedish, but I hope it’s pretty obvious what it is what…) and in Audacity all the devices are the same and the quality is set to 96kHz and 32-bit floating.

When I play a track in Audacity and at the same time start a YouTube Video, I get sound from both sources…???

My understanding was that the application started first (Audacity) got/took/demanded exclusive rights to the soundcard, which should be the Scarlett since all others are not active. Playthrough and overdub are also both unchecked and during recordings I monitor direct on the Scarlett.

This is probably not a problem, when I record (rip) I close down everything but Audacity and when I monitor, manually remove clicks, normalize, set up track labels and so on I never play anything else. It’s just a nagging feeling that IF Audacity doesn’t have total control of the soundcard, there MIGHT be something else using its resources.

I have not yet tried to play a video at the same time as I make a recording, so I don’t know if Audacity has full control in that situation…

Oddly (at least to me) is that Focusrite recommends to uncheck exclusive mode, in order to potentially increase sound quality…? I thought it would obviously be better if the DAW and the soundcard had exclusive rights to each other.

Hopeful that someone can shed some light on this…!
THANKS in advance…!!

Micke

[DVDdoug]

@Micke-T

When I play a track in Audacity and at the same time start a YouTube Video, I get sound from both sources…???

Well… That’s not exclusive.

[EDIT]
If you are recording from YouTube, you are recording/capturing the soundcard/interface output, and yeah… That can be a mix. You should be recording the input… It will work normally if you select the Focusrte as your Recording Device and DON’T select “loopback”.

But when you record you can only record from one source, unless you are using “loopback” or “stereo mix”, etc., which record whatever is coming out of your soundcard/interface, and that can be a mix.

Recording might be in exclusive mode even if playback is not.

I think exclusive mode is supposed to “lock in” the sample rate and it shouldn’t re-sample. With the regular (non-exclusive) settings you can record at 192kHz and if you are using a “cheap soundcard” the drivers will up-sample without telling you. (ASIO drivers also avoid re-sampling but Audacity doesn’t support ASIO.)

You’ll get the actual bit depth from the device (24-bits from the Focusrite) and then Audacity normally converts to 32-bit floating point. (This conversion is lossless). That’s standard with audio editors & DAWs because DSP works better in floating point.

there is a slight tendency towards a “harsh”, or “digital”, sound.

That’s your imagination. …Unless you are trying to go over 0dB and clipping, or unless you are getting dropouts/glitches where the audio gets interrupted. Dropouts are more likely with higher sample rates because there is more data to deal with.

“CD quality” (16-bit/44.1kHz) is generally better than human hearing and WAY-WAY better than analog records or cassettes (which are obviously worse than human hearing). Assuming no “glitches” (and no lossy compression) almost all “quality problems” are on the analog/acoustic side.

[Micke-T]

Hi, and thanks for the super quick reply…!

(we’ll see how this answer will come out, I’m not used to the handling of, for example, quotes…)

If you are recording from YouTube, you are recording/capturing the soundcard/interface output,

I’m only recording from a Turntable → RIAA Preamplifier → Scarlett 2i2 (as an ADC) → Laptop/Audacity.
But are you saying that if I did try to record from YouTube, I would record the output from the Scarlett (through USB) that, when it comes to YouTube, gets the input from the computer (through the same USB cable). In that case it sounds like it would be “easier” for the computer to record what’s already in the computer, instead of using the external card as a “go-between”.

One (of many) things I struggle to get my head around is that digital isn’t a one way thing… It was (is) a LOT easier in the analog domain.

Recording might be in exclusive mode even if playback is not.

That sound reassuring…!! As I said, I haven’t yet tried to record with YouTube running, but I will try, just to try to make sure that the signal from the turntable would be exclusive. In some ways the use of ASIO drivers would make things easier, although I think I understand why it’s “omitted”.

I like to be on the safe side so I keep the bit depth to 24-bits from the Focusrite (and the sampling rate att 96 kHz) and let Audacity converts to 32-bit floating point. I then normalize, and maybe remove some really nasty clicks and pops, before i save to FLAC 24/96 with dithering (although that probably does nothing audible).

That’s your imagination.

No tendencies to go over 0dB, strictly recording with -6 dB before normalizing. BUT (to me) there can be obvious “problems” and “artifacts” that come from nasty USB power, RFI, EMI and lots of other sources, like bad ADCs and DACs and badly implemented software. Lots of this “digital hash” is non-existing in todays systems, but listening to early CDs on a very early portable CD-player (in the late 80s) very often made me feel that I’d rather not listen at all, I had the same thing happening to me recently, with some cabling, hard to define, but it just doesn’t feel good… So for me, keeping every part of the chain reasonably “clean” is good practice. But the majority of todays gear and software for, for instance, streaming, in my case Qobuz through a Bluesound NODE (with updated power supply) into a Hegel H95 DAC/integrated amp, is absolutely wonderful. (BTW The “imagination” was shared, and actually pointed out totally unprovoked, by my girlfriend, equipped with ears that are a LOT more “golden” than mine)

I don’t think it has anything to do with frequency range, signal to noise ratio or measurable distorsion. As you say, the Redbook CD standard is more than sufficient for us humans, but none the less I sometimes detect something, and NO i could not pin it in an A/B or A/B/X test, but after a while when listening it creeps up on me…

But you did put my mind at ease when it comes to what happens in the digital process. I will do some more testing running the laptop on batteries and other small (and cheap) “tweaks”.

THANK again…!!!
Micke

[DVDdoug]

I like to be on the safe side so I keep the bit depth to 24-bits from the Focusrite (and the sampling rate att 96 kHz) and let Audacity converts to 32-bit floating point. I then normalize

Perfect. The only reason NOT to use “high resolution” is if you want smaller files or if you are making a CD and you can always downsample later.

, and maybe remove some really nasty clicks and pops

Audacity has an automatic Click Removal effect, a manual Repair effect, and “worst case” you can “re-draw” the waveform.

I’ve used Wave Repair ($30 USD) for many years. It offers a few different repair methods and it can do an audibly perfect on most (but no all) clicks & pops. Ironically, it seems to work best on the worst defects, maybe because they are easier to find. It’s manual so it’s VERY time consuming but it only “touches” the audio where you identify a defect. (It doesn’t work on high resolution files.)

I also have Wave Corrector which is automatic and now FREE, but I don’t have that much experience with it.

before i save to FLAC 24/96 with dithering

You’re supposed to dither when you downsample so if you are sticking with 24-bits, don’t dither. Plus… Dither is added noise (which is supposed to be preferable to normal quantization noise) and with vinyl, cassettes, or almost anything from analog there is already noise so it’s “self dithered”. Adding a tiny bit of dither-noise won’t make any difference. Plus… At 16-bits you can’t normally hear dither (it’s around -90dB), or the effects or dither, or the lack of dither. It’s even less important at 24- bits where it’s around -140dB.

No tendencies to go over 0dB, strictly recording with -6 dB before normalizing.

The main thing is to avoid clipping which happens at exactly 0dB. Otherwise digital levels are not at all critical. Nothing bad happens when you get “close” to 0dB and nothing bad happens with lower levels, especially at 24-bits where you have tons of dynamic range. Sometimes low levels are an indication of a problem on the analog (or acoustic) side but turning-down the knob doesn’t hurt anything. (Pros often record around -12 to -18dB.)

With analog tape you wanted a hot signal to overcome tape hiss, and tape tends to soft-clip as it starts to saturate over 0dB so it was OK to go occasionally “into the red”. The tape equalization also “softens” clipping/distortion. But with digital… no tape hiss. And digital hard-clips at 0dB.

BUT (to me) there can be obvious “problems” and “artifacts” that come from nasty USB power,

USB power CAN be a problem. The digital data is immune to noise but noise can get into the interface’s analog electronics. It’s usually a high-pitch whine from the switching power supply and the digital switching inside the computer. (You can get it even with battery power because there are switch-mode voltage regulators in your laptop and of course lots of high-speed digital data moving around.) But that tends to more of a problem with the microphone preamps because any noise gets amplified. It’s rarely an issue with line level inputs/outputs. And I’d expect your Focusrite to be pretty-well filtered.

[Micke-T]

Hi again DVDdoug (and any others reading this…)

Sorry about the delay (or maybe “latency”… ) in responding, lots of other stuff that needed to be done and also a LOT more searching, reading and testing.

Audacity has an automatic Click Removal effect, a manual Repair effect, and “worst case” you can “re-draw” the waveform.

I tried the automatic Click Removal a while ago and also to manually “re-draw” the waveform, but never the Repair effect, I will look into that and maybe also Wave Repair, but when you say that Wave Repair doesn’t work on high resolution files, I guess that would include 24-bit/96 kHz…? In that case I probably stick with Audacity’s built in effects. But I’ll also check out Wave Corrector, thanks for the suggestions!

…if you are sticking with 24-bits, don’t dither.

I thought that since Audacity internally works with 32-bit floating, going back down to 24 bit for export was considered “downsampling”… And I also did it just to be on the safe side, I found a quote from a famous Sound (Mastering…?) Engineer that said “When in doubt, dither”

The digital data is immune to noise but noise can get into the interface’s analog electronics.

Yes it can, and it’s not only the obvious things like “hum” and “whine”, it can also affect the ADC’s (or DAC’s) performance in converting to/from digital to analog. The same thing with the choice of Host, according to many WASAPI sounds clearly better than, for instance, Windows Direct Sound. The latter actually measures really bad with a huge increase in distortion and jitter. But WASAPI is only clearly superior when it’s used in Exclusive Mode. So IF the Exclusive Mode isn’t recognized, that could be a reason for the slight tendency to “harshness” that I “believe” I hear.
Unfortunately, all the information I found refers to playback from different Media Player’s, but when I play back a piece I’ve just recorded, Audacity acts like a Media Player, and that should be handled exclusively.

When using exclusive mode in WASAPI and setting the “Media Player” to use WASAPI, with WASAPI set to be in exclusive mode, the volume controls in Windows should not work when playing through the Media Player (in this case playing back a file in Audacity). But whatever settings I use I can still use the volume controls in Windows. This “should” mean that the Windows Mixer is being used, which it should NOT be if the software is using WASAPI with WASAPI set to exclusive.

So, to me, there is NO way that exclusive mode is being used when playing back files…!
So, how can I be SURE that it’s exclusive when recording?

To test recording with different settings -

1) In Windows (under Sound Devices) I set the Computer’s Internal Soundcard to be the Default, at 24/48 (the highest possible for the internal card) and Exclusive Mode turned off. For the Focusrite I set WASAPI to be in Exclusive Mode. In Audacity I set WASAPI as Host and as Recording Device (NOT loopback!). I started recording from Vinyl and then started to play a YouTube video, during the already started recording, I got the video’s sound out of the Computer’s Headphone Output and nothing more than the Vinyl through the Focusrite.
So far so good!

2) I then set the Focusrite as Default player in Windows with everything else the same as in 1). When playing a video now (together with an already running recording in Audacity) I get the video’s sound output through the Focusrite, mixed in with the signal from the Vinyl. This is listening with earphones plugged in to the Focusrite and doing Direct Monitoring of the input from the turntable/RIAA preamp.
But the sound from the video does NOT end up being recorded in Audacity…!!
Well, that’s very good, but it still doesn’t sound very exclusive…! And although it doesn’t end up in the recording there is still lots of data being transported TO the Focusrite over USB (the video’s sound), at the same time as there is the important data (the digitized vinyl recording) traveling FROM the Focusrite to the Computer over USB. Potentially taking up resourced in both ends and increasing the risk of interfering with the ADC/DAC.

In the above two scenarios there was NO difference with WASAPI set up in Exclusive, or Non Exclusive, Mode. Nothing but the digitized analog input from Focusrite gets recorded either way. But I’m not really sure what that means when it comes to the “exclusivity”…?!?

In Exclusive Mode it shouldn’t matter that the sampling frequency isn’t the same… But maybe it does. But I don’t want to use 48 kHz for my Vinyl Rip’s.

Naturally, the obviously easiest and most natural thing to do is to -
A) Set the internal sound card to be the Default player with no exclusive rights, and also mute it.
B) Set WASAPI to be in Exclusive mode for the Focusrite.
C) Never play anything else, or even do anything else, on the computer when recording. That is of course good practice anyway, to let Audacity have all available resources in processor, memory and disk writing.

BUT it bugs me like … not to be 100 percent sure that there are no other programs, or devices, being stand by or even trying to communicate with the external sound card (the Focusrite) or Audacity or any other resources…!

I think I’ll approach Focusrite with some questions to see if there might be something in their driver that’s got something to do with all this. They do after all, oddly enough, suggest to turn of Exclusive Mode in WASAPI, when everybody else, including Audacity says that it’s essential to use it in that mode to get the best quality. From the manual: enabling both “Exclusive Mode” boxes in Windows Sound allows Audacity to request audio direct from the device without resampling. So why do Focusrite recommend you to turn it of to increase sound quality…?

Thanks for all your input…!!! My search continues…
Micke

Done two small edits to clarify… M