Plugins and Upsampling. Wrong order

Try using ‘Power of Two’ upsampling strategy… You gain nothing by using your ‘Custom’ upsampling strategy… You need to eliminate this variable of the Custom upsampling settings, so to get a better grasp on the potentials that may be playing into the behavior you are experiencing.

:notes: :eye: :headphones: :eye: :notes:

Same result. Btw. 768 is pow 2 of 192 or 48

Okay, Yes, I know the results of Power of Two will either be 705.6kHz or 768kHz, the logical maximum results that your DAC will support.

What does Dirac Support say about this behavior?

Hi @Ralf_Napierski,

I looked at the thread and tried to reproduce the issue on our end with success. Can you please contact DIRAC if their plugin has some sort of logs so we can see the sample rate it detects?

1 Like

You open some questions here… Were you able to reproduce the issue or not? It also seems you are alluding to an operational scenario with this plug-in, where the signal path is being routed post upsampling with this statement:

“Can you please contact DIRAC if their plugin has some sort of logs so we can see the sample rate it detects?”

Is this uniquely associated to this plug-in or other convolution type plug-ins?

:notes: :eye: :headphones: :eye: :notes:

I have been able to reproduce the issue.

I don’t know, I need to cehck this with others convolution plug-ins.

2 Likes

Hi @Antoine,
thank you for your info.
I have contacted DIRAC. But no answer until now.

Hi @Antoine, there is still no solution for the problem?

Works as expected with the HangLoose Convolver for Room Correction DSP and Michelangelo EQ before Audirvana’s r8Brain upscale…

2 Likes

Thanks @Ddude003

I am not sure what Audirvana’s expected behavior is as the r8brain “upsample” is being communicated to HLC even though it comes “after” the plugin…

Here is an example. I have set r8brain resampler to X2 only. I am playing a 44.1 kHz tune, and processing is showing being processed at 44.1 kHz, but in HLC shows processing at 88.2 kHz:

Setting r8brain resampler to max PCM frequency which is 768 kHz, Audirvana processing still shows 44.1 kHz, but looking at HLC, the max sample rate is being communicated instead, so HLC looks for a matching sample rate filter and if it can’t find one, it takes the highest sample rate filter it can find (in this case 352.8 kHz) and then HLC resamples the filter (also using r8brain resampler built into HLC) to match the Host sample rate:

The Topping E30 DAC I use also supports DSD, but the convolver does not, so even though Audirvana shows still processing at 44.1 kHz, convolution is bypassed as it knows nothing about DSD.

From my end, it looks like Audirvana’s r8brain resampler is communicating it’s sample rate to the plugins instead of the source sample rate from the music being played.

I believe this is the same issue with using Dirac Live Processor plugin that @Ralf_Napierski originally reported.

1 Like

It is not the r8brain that is the issue here, it is how audirvana implement upsmapling . If you use Sox as upsmalper in Audirvana , the same problem is there.

@Mitchco @koldby
I personally modulate all PCM to DSD128 using r8Brain with 112dB’s ‘Redline Monitor’ HRTF virtualization AU plug-in processing, and Audirvana displays the processing sample-rate and output sample-rate as expected… The same is true with SoX…
:notes: :eye: :headphones: :eye: :notes:

@koldby Yes, for SoX too. It does appear that HLC is receiving the upsampler sample rate and not the source sample rate. This is something that I cannot change as the Host (Audirvana) is responsible for sending sample rate to the plugin.

Hi @Agoldnear does the Redline Monitor plugin show the internal sample rate in its display? I don’t see it in it’s GUI…

What I am saying is if you look at HLC screen shots above, while Audirvana is “displaying” 44.1 kHz “Processing” the actual sample rate HLC is receiving is from the upsampler, and not the music source sample rate.

1 Like

No… Redline Monitor is a 64bit application good to 384kHz… My DAC is seeing 1-bit 5.6MHz from Audirvana…

Note: @Mitchco It’s interesting that the input/pre-load buffer read-out in your screenshots are showing the FLAC file bit-depth changing from 16/44.1 to PCM 64/44.1… Something to do with the FLAC → PCM conversion processing? Compare this with my screenshot.

Here is Windows plugin example using https://juce.com/ As the site says, the most widely used audio framework for developing audio apps and plugins (including HLC and Dirac Live Processor for example).

Before any audio is played, the Host (i.e. Audirvana) communicates to the plugin the sample rate through a function called prepareToPlay

The SampleRateDisplay plugin: https://accuratesound.blob.core.windows.net/sotadrc/SampleRateViewer.vst3.zip simply displays the prepareToPlay sample rate passed in to the plugin from Audirvana.

To use, download and unzip. Manually copy the SampleRateViewer.vst3 folder to: C:\Program Files\Common Files\VST3

Launch Audirvana and load the plugin, turn real time mode on, For now, leave upsampling off.

Now press play on a variety of tunes and you will see that whatever music sample rate is being played is being passed to the plugin and displayed in the SampleRateViewer:

Engage the resampler. In the SampleRateViewer, whatever upsampler settings have been set are now also being sent to the plugin. Here I played the same song again, but with 2X upsampling and that is also reflected in the sample rate passed into the plugin:

So while the source material is 88.2 kHz the 2X resampler setting is what is being sent to the plugin.

Whether this is expected behavior or not is for @Antoine to say.

1 Like

Hi guys, I got the confirmation that when you are playing a track with a plugin in realtime, upsamling is applied BEFORE the plugin processing. I’m very sorry, it’s really confusing in the app.

Which would make sense… In my non-techie opinion.

Is this just in the context of these types of convolver plug-ins? Because how can this be when a PCM file is being upsampled/modulated to DSD and the plug-in is PCM based?

In the context of Real time use of plugins with upsampling activated in general, not only with convolver plugins.

So, because no real-time control is available during upsampling to DSD then we can expect the sequence to be logical… Plug-in module → Upsampling/modulation ?