Kernel Streaming, ASIO, WASAPI... and music players (Foobar, JRiver...)

[bennetng]

If the difference is 2Vrms vs 1Vrms, then keeping foobar's volume control at -6dB should keep the output to not go over 1Vrms.

ReplayGain will further decrease individual files' volume based on the stored metadata.

 

[zermak]

I've tried WASAPI shared for the two audio files. To my hears it seems like the a.mp3 file doesn't clip anymore at around 45% or 46% Windows volume and indeed it is the -12 setting in the Khadas (TUSBAudio) control panel.
What to do then? I mean, is the clipping a real issue in the Windows platform? Will I encounter games or programs that will suffer from it? You made that clips on purpose to validate this problem but I hope software houses and so on use some brain (and have technical knowledge) to avoid it.
I am not concerned about the music because of ReplayGain.

If the difference is 2Vrms vs 1Vrms, then keeping foobar's volume control at -6dB should keep the output to not go over 1Vrms.

ReplayGain will further decrease individual files' volume based on the stored metadata.

Ok thank you, and is it the same to add a preamp gain of -6dB when using ReplayGain? And what if I want to rescan all my music and use a gain in which the loudest file has this 1Vrms value? Now ReplayGain is at the default 89dB, should I use a 83dB value (or is this another scale)?

And what about on the whole Windows system. Should I use Windows volume? And to what percentage? Or maybe it is better to use the MASTER in the TUSBAudio control panel?

Lots of questions
Thank you for your time and the future answers

 

[bennetng]

It seems to me that you need a universal solution on this 1Vrms vs 2Vrms issue, not only on foobar, but other apps, games... as you mentioned.

I am going to change the previous suggestion that using -6dB on foobar, instead, use that -6dB on the Khadas's volume control as it is system wide. In this way nothing in your system can go over 1Vrms. In the case of foobar, you can continue to use ReplayGain to make sure the audio data are not internally clipped before leaving foobar.

Games and generic apps (like browsers) don't use exclusive mode so they are unlikely to clip in this way, the limiter should be able to prevent hard clipping to happen. If the games have option menu to adjust volume of dialogs, sound effects and music then you can try to adjust them as well.

 

[zermak]

It seems to me that you need a universal solution on this 1Vrms vs 2Vrms issue, not only on foobar, but other apps, games... as you mentioned.

I am going to change the previous suggestion that using -6dB on foobar, instead, use that -6dB on the Khadas's volume control as it is system wide. In this way nothing in your system can go over 1Vrms. In the case of foobar, you can continue to use ReplayGain to make sure the audio data are not internally clipped before leaving foobar.

Yes, universal for now because I am using the Khadas on my Windows PC but I plan to get another (maybe the fixed hump version) for the living room amp.

I will go with the -6 in the MASTER volume then. Little side note, this slider has a 0 (full) to -127 (muted) values (a 7bit int?) and I am not sure if these are really dB (it has the 0dB icon only on the bottom) or if it is something else; any ways to discover it?

Games and generic apps (like browsers) don't use exclusive mode so they are unlikely to clip in this way, the limiter should be able to prevent hard clipping to happen. If the games have option menu to adjust volume of dialogs, sound effects and music then you can try to adjust them as well.

I already control their volumes cause most games are too loud with my usual analog volume settings and it is very annoying XD

 

[maty]

Without wishing to enter into controversy, I tell you:

Test with an excepcional vinyl rip, with a lot of detail, because the cartridge I think.

Elton John - Goodbye Yellow Brick Road (1973), Vinyl x2, EMI-Odeon / Zafiro, Spain

https://www.diyaudio.com/forums/music/6958-playing-listening-post5821248.html

A2 Candle In The Wind

Just now, I have been alternating Kernel Streaming and Wasapi in JRMC v24 64 bits and I notice the difference clearly, especially with the piano when I alternate. With KS the sound is more detailed. Wasapi has more LF.

A3 Bennie And The Jets

The same. It is very evident.

These last days I have an excess of detail, so I have measured and I had 239 Vac. With KS I have audio fatigue with GYBR. With high voltage I prefer Wasapi. I had and have 239 Vac

 

[bennetng]

I will go with the -6 in the MASTER volume then. Little side note, this slider has a 0 (full) to -127 (muted) values (a 7bit int?) and I am not sure if these are really dB (it has the 0dB icon only on the bottom) or if it is something else; any ways to discover it?

Unfortunately I don't use Khadas so you may ask other members.

 

[maty]

ASIO4ALL v2.13
Buffer: 256 samples. In and Out: 32 samples. Kernel Buffers: 2.

Less detailed, more LF.

Better with 384 samples. 320 samples.

 

[zermak]

Without wishing to enter into controversy, I tell you:

Test with an excepcional vinyl rip, with a lot of detail, because the cartridge I think.

Elton John - Goodbye Yellow Brick Road (1973), Vinyl x2, EMI-Odeon / Zafiro, Spain

https://www.diyaudio.com/forums/music/6958-playing-listening-post5821248.html

A2 Candle In The Wind

Just now, I have been alternating Kernel Streaming and Wasapi in JRMC v24 64 bits and I notice the difference clearly, especially with the piano when I alternate. With KS the sound is more detailed. Wasapi has more LF.

A3 Bennie And The Jets

The same. It is very evident.

These last days I have an excess of detail, so I have measured and I had 239 Vac. With KS I have audio fatigue with GYBR. With high voltage I prefer Wasapi. I had and have 239 Vac

I don't like Vinyl rips. It has been validated that from the same master a vinyl rip has a (fake) higher DR than the CD copy and the differebce is mostly becasue of the obvious parts into game to rip it (cartridge, arm, ADC?, block and so on). Unless the material is only available on vinyl then a rip is convinient but in this modern era where all pratically is from digital (and if the two masters for both vinyl and CD producions are the same) then I prefer CD rips.

If I am not mistaken WASAPI uses the KS lower layer to interact with the hardware so that's kinda weird.
And how the Vac has something to do with it considering (I suppose) that most of the electronic equipments has some filters/power stage before going into their own hardware?
In my PC I have a decent PSU (with low ripple values and so on) and before going into that I have a UPS with a pure sinusiodal wave output (Vac now is around 222).

Unfortunately I don't use Khadas so you may ask other members.

Ok thank you anyway. I'll find the answer somewhere else I appreciate your time.

 

[maty]

One of my improvements, the first, was to upload the bias of the amplifier. I suppose that with the excess of voltage the sound changes, maybe it causes the sum of the two factors an increase of the third harmonic H3.

Most of the time the value of the DR coincides, in the good recordings that have not been remastered. I listen to music in all kinds of formats except streaming. Yesterday, DSD128.

Eric Bibb - A Selection of analogue Eric Bibb (2013) (Opus 3) [DSD128]

https://www.diyaudio.com/forums/music/6958-playing-listening-post5826271.html

You have a preconceived idea about vinyls. I guess most of them sound that bad but not the ones I hear. Ripping vinyl is an art!

 

[zermak]

I am just a realistic and objective (as mush as I can be) person and I base my assessments by facts.
Vinyl has its fashion but as I stated (and the DR issue confirmed by people working in the field) if you have the same master fot both CD or vinyl production I choose the CD version because I am sure that I get (bit by bit) what the master engineer wanted to achieve and I don't have to fight with what a cartridge, arm, base, a possible defected vinyl and so on can add to the sound before reaching the amp.

Well you can indeed say that ripping a vinyl is an art considering all the factor involved and as it is, it makes the original (which is already modified by implementing it on analog platform) something else ;P
And I am not talking about liking it or not but on a technical point of view because I am pretty sure I would enjoy some vinyl audio sessions regardless

 

[maty]

So we understand each other: I look for the best sound and that excites me, I do not care about the format. Almost always on vinyl because I listen to recordings from analog masters before 1980, with acoustic instruments. Today technology is much better but it is being used very badly, with exceptions.

Better not deviate more from the thread.

 

[zermak]

So we understand each other: I look for the best sound and that excites me, I do not care about the format. Almost always on vinyl because I listen to recordings from analog masters before 1980, with acoustic instruments (mostly) electroacoustic transducers. Today technology is much better but it is being used very badly, with exceptions.

Better not deviate more from the thread.

To each his own

PS: I have corrected your statement

 

[BYRTT]

...I will go with the -6 in the MASTER volume then. Little side note, this slider has a 0 (full) to -127 (muted) values (a 7bit int?) and I am not sure if these are really dB (it has the 0dB icon only on the bottom) or if it is something else; any ways to discover it?...

Unfortunately I don't use Khadas so you may ask other members.

Have a Khadas TB in AUDIOPHONICS cheap box for my head phone setup, OS is Win7 and Thesycon XMOS driver v2.24, player is JRiver v24 playing a 100 Hz tone -20dB and all my HD650 DSP settings is disabled except a 20dB volume boost so that JRiver report peak level is at 99%, specs for DMM used is true RMS up to 500Hz.

LEFT CHANNEL 0dB:

LEFT CHANNEL -6dB:

LEFT CHANNEL -7dB:

RIGHT CHANNEL 0dB:

RIGHT CHANNEL -6dB:

RIGHT CHANNEL -7dB:

 

[zermak]

@BYRTT Damn fast! I just opened a topic on Khadas' forum XD and well thank you!
If I am correct the output from JRiver then is back a 0dBFS? I will use -7 then (maybe even -8 considering the stupid loud games and I like the binary stuff ahah). And so the slider is correctly showing decibels if my understanding of RMS and dBv is correct (source): in your case you indeed need a little more than -7dBV to have 1Vrms.

 

[bennetng]

I have an old post about ReplayGain modes during playback in older versions of foobar2000, but the quoted part applies to the latest version as well.

ReplayGain is a widely used standard in various audio players (check Wikipedia). It stores Track peak and Album peak as metadata in audio files. Track peak is the peak of individual audio file and album peak is the highest peak of an album. In this way users can maintain the relative loudness of an album during playback.

The "Preamp" section sets a target loudness in terms of LU. By default foobar2000 uses -18LUFS as target but as you can see, there are sliders for you to adjust. With the "prevent clipping according to peak" option selected, even if you set the slider all the way to the right, files with RG info won't clip, but in this way the target loudness cannot be maintained.

 

[bennetng]

To demonstrate how ReplayGain recovers peaks, let's use the example song I mentioned in Archimago's blog:

There is a way to grab the audio stream from Youtube directly (disclaimer: I am not a lawyer, it is just a techincal demo), the underlying format is a 48kHz Opus file. Right click the Youtube video, choose Stats for nerds will show the loudness normalization value (-2.3dB)

Here is a 24-bit digital recording when using the Youtube player:

In foobar2000, enter the same -2.3dB value:

Here is a 24-bit digital recording from foobar2000's WASAPI shared output, as we can see the compressed peaks are recovered.

Here is the null test result between two recordings revealing the recovered peaks:

 

[zermak]

So is it better to use ReplayGain with the prevent clipping according to clip peak option selected? And in this case even if the loudness will be lower (in case of clipping tracks) the DR of the track will be mainteined? Correct?

Just for info since I use random playback I prefer to use the track source in ReplayGain.

EDIT: corrected

 

[bennetng]

So is it better to use ReplayGain with the prevent clipping according to clip option selected? And in this case even if the loudness will be lower (in case of clipping tracks) the DR of the track will be mainteined? Correct?

Just for info since I use random playback I prefer to use the track source in ReplayGain.

Yes, this mode suits your playback habit.

 

[maty]

Just now I am listening to an exceptional 24/192 vinyl rip of Dire Straits - Dire Straits (1978, UK) and the differences between Kernel Streaming, Wasapi and ASIO4ALL have been enlarged. JRiver MC v24.0.78 64 bits.

https://www.discogs.com/Dire-Straits-Dire-Straits/release/1308001

Since I do not have measurements that show it, you only have to trust my word.

The key: the new and exceptional good cartridge.

 

[Veri]

Since I do not have measurements that show it, you only have to trust my word.

I think KS, Wasapi and Asio all sound the same