Fixing Foobar2000 problems with ASIO2

[escape2]

For years now, I have been experiencing random freezes in Foobar2000, causing "Unrecoverable playback error: Timeout" when advancing into a later section of a track/fast forwarding.

I finally came across this old thread the other day that describes my problem and why it's happening:
https://hydrogenaud.io/index.php?topic=95031.0

So I've installed ASIO2 from here:
https://sourceforge.net/projects/foobar2000-wasap2-output/

I'm happy to report this solved my problem. However, I am not too clear on the differences between the original ASIO plugin and ASIO2. Just wondering if there are any downsides to ASIO2 that I should be aware of. Anyone knows?

Thanks!

 

[escape2]

OK, so I've indentified first problem with ASIO2 - gapless playback is no longer gapless.

 

[Propheticus]

You could also try the official Asio component, but play with the buffer settings.
Something safe could be 1500ms buffer in Foobar, 1024 samples in the Asio driver settings for 44.1kHz playback (2048 for >44/<96kHz and 4096 for >96kHz). Ticking 'Safe mode' will even increase the output buffers more if you encounter stability issues.

Alternatively: do not use Asio. Chances are you won't hear a difference using the Windows DirectSound route. Downside is non-exclusiveness, and resampling (or constant switching of audio settings).
If you're concerned about Windows doing the sample rate conversion for you, you can use the 'Resampler (SoX)' component in F2K to convert everything to 44.1kHz (default settings are best left alone).

 

[escape2]

You could also try the official Asio component, but play with the buffer settings.
Something safe could be 1500ms buffer in Foobar, 1024 samples in the Asio driver settings for 44.1kHz playback (2048 for >44/<96kHz and 4096 for >96kHz). Ticking 'Safe mode' will even increase the output buffers more if you encounter stability issues.

Tried all of those already - did not fix the problem.

I have now switched to WASAPI (Event).

 

[Rendboell]

I have gapless playback after removing the "Override Replaygain" component.

In my experience the ASIO2 and WASAPI2 plugins are improvements to the official Asio and Wasapi components.

It's not a big difference, but it is there.

I actually discovered the difference when playing audio only with MPC-HC and the Directshow Multichannel Asio Renderer".

Sound quality has improved all over the frequency spectrum with greater sound stage and high and low end clarity and details. The stereo perspective is "bigger" with voices and instuments placed distinctly in 3D across the stereo perspective.

I have had a few Foobar2000 crashes, but PCM, SACD and DVD-Audio playback is now stable after some tweaking in Foobar2000 preferences "Advanced Settings". In my case installing the patch (see below), unticking SIMD Instructions and Visualisations, as well as setting "end point buffer sizes" according to hardware specs.

The Asio2 and Wasapi2 components can be found here:

https://sourceforge.net/projects/foobar2000-wasap2-output/

If problems with sample rate change, then try the ASIO2 patch from Ticket #12:

https://sourceforge.net/p/foobar2000-wasap2-output/tickets/12/

Best regards,

Jens

 

[Propheticus]

Unless something was seriously broken, this "sound quality has improved" sounds very unlikely. Because Asio (in itself) is transparent -or bit-perfect if you like- , there should be exactly 0,0 difference in SQ between non-broken implementations. What can chance is stability and efficiency. Or there are more hidden DSP features in the alternative plugin other than pure transport.
Differences between broken and not broken are very obvious like popping, clicking, cutting out* or crashing altogether. Not improved soundstage and other audiophile prose.

So either this plugin fixes a soundcard (driver) or other hardware incompatibility for you, applies processing, or you're imagining improvements. Until I'm shown measurements with proof of differences, I'll assume the latter.

Note the comment by Peter (author of FB2K) on the use of Asio: "...this component is meant for systems where ASIO is the only available output method. It is highly recommended to use the default output modes instead of ASIO. Contrary to popular "audiophile" claims, there are NO benefits from using ASIO as far as music playback quality is concerned, while bugs in ASIO drivers may severely degrade the performance. "

The only reason I use Asio, is to prevent Windows resampling high sample rate music before it reaches the DAC. I honestly can't hear a difference when using a SoX resampler to 44.1kHz and outputting via DirectSound though.

*these symptoms are mainly buffer (underrun) issues. The default buffers sizes per sampling rate can be wrong or set too low by the user; simply increase buffers. Another cause being an inefficient design and/or not so powerful PC causing unsolvable underruns (other than: run less stuff).

 

[pLudio]

Just use WASAPI in Exclusive mode if you don't really need ASIO.

https://foobar2000.org/components/view/foo_out_wasapi

 

[Rendboell]

I am sure I hear a difference, subtle, but it is there.

There is some kind of "musical feeling" about it, if you can express it that way (I am not a native english speaker).

And I don't think my implementation is broken, as it is there with different kind of inputs/implementations.

Until now, I have allways been a little disappointed with the sound quality of the official Foobar2000 Asio and Wasapi.

Initially, I noticed the difference with MPC-HC + the Multichannel Directshow Asio Rendererer, as reported here by MZ in post 5: https://sourceforge.net/p/foobar2000-wasap2-output/tickets/12/

I hear the difference with various implementations to the RME Fireface 800:

(1) Input from desktop PC with Foobar2000 + Asio2 (Firewire) and spdif input from a laptop with Foobar2000 + Wasapi2 (optical).
(2) Firewire input from MPC-HC with Multichannel Directshow Asio Renderer (http://blog.familie-buchberger.at/category/projects/).
(2) Direct spdif inputs with Denon (optical), B&O (coax) and Samsung TV (optical).

All with better sound quality compared to the official Foobar2000 Asio and Wasapi components. Audio is lossless wav/pcm and the same audio master is used as input in all implementations.

Actually, now I prefer the inputs from Asio2, Wasapi2 and Multichannel Asio over the direct inputs. So my ranking is:

(1) Inputs from desktop PC (Asio2), laptop (Wasapi2) and MPC-HC (Multichannel Asio).
(2) Direct inputs with Denon, B&O and Samsung TV (optical, coax).
(3) Inputs from desktop PC (official Asio) and laptop (official Wasapi).

Of course, it could be a bug in the RME Asio/Wasapi driver implementations, but I don't think so, as it is used with several professional RME audio interfaces.

It could also be that the RME Fireface 800 is rather old hardware, but still the difference is there.

The latest Asio2 release is from 2017, and the developer writes that it is build upon the Steinberg Asio SDK v2.3, and I think a more recent build than the official Asio from 2012? With Wasapi2 I don't know, but it could be build upon the same base?

I am no audio expert, so I should not make any judgement on implementations. Also, I can not make any measurements. So it's a pure subjective view. I am just a regular guy, who loves listening to music.

Try it out, it is pretty easy to install the plugin and hear for yourself.

https://sourceforge.net/projects/foobar2000-wasap2-output/

Use the Asio2 patch, if you have problems with sample rate change (see my post above and here: https://sourceforge.net/p/foobar2000-wasap2-output/tickets/12/).

Best regards,

Jens

 

[Veri]

Jens,

This isn't really the forum where people go by other's "musical feelings". ASIO and WASAPI exclusive modes should be perfectly bitperfect. If such changes in sound would be true, they'd not be bitperfect. A condradiction in itself..

 

[Rendboell]

Veri,

But then I don't understand why there are differences between my "bitperfect" implementations?

Still I am no expert, but could it be that bitperfect implementations are not "strictly" transparent in hardware nor in software with various hardware components in various digital/analog signal paths/inputs, as well as influence from hardware and software environments on signal paths in computers?

I don't think you would disagree, if I say that there are differences in both the analog, as well as in the digital domain dependent on implementations?

Best,

Jens

 

[Propheticus]

Like I said, they differ in performance (as in: require less processing power), stability and compatibility. Just not in sound quality, if not broken.
Putting it bluntly, the sound quality differences are probably between your ears. I'll not rehash things discussed many times before and which have been explained better than I could. Keywords: placebo effect, expectation bias, confirmation bias.

What you're saying is a contradiction. If both are transparent / bit-perfect, there is 0 difference.
Analog has nothing to do with it. This is all in the digital domain. ASIO/WASAPI are merely interfaces to get digital sound data from an application to the soundcard (driver). It's the soundcard or external DAC that does the conversion into analog.

 

[pLudio]

Record the different output options with a virtual audio cable or S/PDIF output and compare with DeltaWave.

 

[Rendboell]

pLudio,

Thank you for your sugestion

I will look into it tomorrow.

Veri,

The things I am talking about is in the digital domain, I was just referring to analog for comparison. It's the same hardware, RME Fireface 800, that gets 6 different "bitperfect" digital inputs, but still there is a difference.

What we are talking about here is transparency of an Asio plugin released in 2012 along with Windows 8.

I could understand if one of my implementations were broken, but not all 6 of them. Others are reporting the same with far better hardware than mine (RME ADI2DAC).

Over the years, I have come across quite af few implementations with Asio and Wasapi that sounded terrible with distortion and jitter all over the place.

In the mean time there has been a lot of improvements in windows audio, and I agree that we now are talking minor differences between audio drivers.

But I guess that you really believe what you are saying, so I challenge you to try it. If you have Foobar2000 installed, it takes just a few minutes to install, like any other component.

In my experience, you get the biggest improvement with big high quality speakers (not headphones), most significantly in soundstage and imaging, as well as low and high end clarity and details. Instruments are coming out in the open, so to speak.

Best regards,

Jens

 

[AnalogSteph]

If there are any real audible differences whatsoever between ASIO and WASAPI (exclusive mode) output, you have a broken sound driver. Or at least one that doesn't do what it says on the tin. I have come across so many driver issues in the last two decades that I would never rule out the possibility completely.

If I had any serious doubts in the matter, I would generate some test tones (Audacity is fine) and break out the multimeter to check voltage output. Maybe one of the output formats passes through some digital attenuation (neither should), so levels don't match. That would immediately sound different. If you don't have a moldymeter (a half-decent one is cheap enough, it may just not be very accurate above 400 Hz or so), setting up a loopback test and recording your test tones with both output drivers may be worth a shot. RMAA can also save its level and test tones, so you can do an "offline" analysis, even from a recording.

Also watch what format the hardware is actually being set to - I've seen cases like the old Asus Xonars where the only way to actually change hardware settings was the built-in ASIO driver, and that one proved rather broken (severe dropouts when all cores loaded with no way to get rid of them), so I ultimately settled on shared mode with settings managed via XonarSwitch. I think it also cost me 3 dB of level. Those cards actually were a real can of worms - I had to go to an older driver build with said broken ASIO driver before the hardware would be initialized correctly beyond 48 kHz (mind you, the last official driver version was one that worked properly, but I'm not sure that same would apply on a newer Windows version). I think the Linux guys are much happier with these overall...

 

[bennetng]

Perhaps read what MC_RME said:
https://www.audiosciencereview.com/forum/index.php?threads/foobar2000-modular-freeware-audioplayer-for-windows-developed-by-peter-pawłowski.6452/post-220947

In fact, I was surprised (and impressed) that the HA admin took my crash report so seriously even when those are unofficial third party components.
https://hydrogenaud.io/index.php?topic=111980.msg923559#msg923559

For me, the official ASIO plugin (from Peter) worked flawlessly, the alternative one actually caused issues.

 

[Rendboell]

AnalogSteph, Veri, bennetng, Propheticus, pLudio

Thanks for your responses

I was actually just responding to OP's problem with gapless playback and wanted to share my experience with Foobar2000 + Asio2 compared to the official Asio.

But as always in a serious and professional forum like this, it creates responses about good or bad, which is what this forum is all about

I have had stability issues and broken audio drivers, but no more with any of my implementations, Asio2 is stable (as was the official Asio before), running all day with a different mix of sample rates going from CD to HDtracks, DVD-Audio and SACD.

The Asio2 is still beta2 and not always just plug&play, but requires you to adjust settings according to hardware (buffer size, SIMD, etc.), including if needed, removing components (gapless playback) and installing patches (sample rate issue).

Look here for the developers comments on the plugin https://hydrogenaud.io/index.php?topic=111980.msg924178#msg924178

As with all kinds of hardware and software, implementations are different, and some are better or maybe just different to others, and this also applies to Asio and Wasapi implementations, which Matthias Carstens of RME Audio is pointing out (as referred to by bennetng).

If I get some spare time, I will compare outputs with DeltaWave, as suggested by pLudio.

I recommend you to try the Asio2, it is not a big hazzle, and if you have problems, then report them here and I will do my best to point to possible solutions.

Best regards,

Jens

P.S. the Wasapi2 plugin has to be installed separately. Only Asio2 is installed if you install the downloaded zip package. You have to split the zip package in two separate zip packages to get Wasapi2 installed, or manually copy separately to the components folder (C:\Users\"user"\AppData\Roaming\foobar2000\user-components). If you play both 16 and 24 bit audio, you have to set the Wasabi2 to output 32 bit in the output preferences to make it work.

 

[Propheticus]

Since the 'normal' Wasapi (exclusive) and Asio components just work, there's no need to try something else. Let alone unstable betas.
It won't improve sound quality, needs adjusting and tweaking to work and is even a step back in functionality.

I get it, you like to tinker. And because you invested a lot of time tinkering you want to (unconsciously) believe your efforts have made a difference.
Nothing wrong with a hobby and tinkering. But for my daily playback I'm not fixing what ain't broken.

 

[escape2]

OP here... thanks for all the comments.

As it stands now, I'm happy with WASAPI (Event) in Foobar2000. No more "Unrecoverable playback error: Timeout" when advancing into a later section of a track/fast forwarding, and gapless playback works fine. So for now, I see no reason to experiment with anything else.

One thing I'm not totally clear on is differences between WASAPI (event), (push), and (shared), and when to use which.

FYI, I'm using buffer length of 500ms, and my output format is 32-bit. My DAC is Loxjie D30.

 

[Rendboell]

escape2,

You're welcome

Wasapi (event) means that the sound driver is "pulling" the audio samples directly from the audio files, while Wasapi (push) means that the software player is "pushing" the audio samples to the sound driver.

It should not matter sound wise, but I think event should be prefered.

I have no experience with Wasapi (shared). It should be non-exclusive, meaning that you will be able to play audio simultaniously from various sources, foobar, youtube, facebook, etc.

Look here for further reading: https://www.thewelltemperedcomputer.com/KB/WASAPI.htm

Best regards,

Jens

 

[escape2]

escape2,
I have no experience with Wasapi (shared). It should be non-exclusive, meaning that you will be able to play audio simultaniously from various sources, foobar, youtube, facebook, etc.

Look here for further reading: https://www.thewelltemperedcomputer.com/KB/WASAPI.htm

Yes, the downside to shared mode is that some audio file content played through Foobar2000 will be subject to rate conversion if it differs from whatever the Windows audio panel is set to.