Audirvana Origin

[Mark Robinson]

On 5/3/2022 at 10:22 AM, The Computer Audiophile said:

I use Audirvana Studio with Ravenna and can report it's rock solid. 

Hi, I use the Merging Hapi mk2 as a mastering engineer and would like to get Audirvana streaming to it via Ravenna but I cannot find a setup guide anywhere. 
 

Is there a resource you can point me to or instructions on how to get Audirvana’s AES67 I/O sinks posted on the network for me route from?

 

Thanks for any help you can give. 

[Mark Robinson]

1 hour ago, The Computer Audiophile said:

Hi Mark, I’ve been using the Anubis and received my HAPI MkII yesterday. I have both working with Audirvana. I’m sending 7.1.4 12 channel Atmos content from Audirvana to them.
 

Which operating system are you using and how many channels? 

Im using Mac OS 12.3.1 on an intel and Audirvana Studio 1.12.2.

 

7.1.4 over AES67 is impressive. All I need is a single stereo feed.  

 

I have been using the Merging Core Audio driver with Audirvana np, but Id like to bypass CA with Ravenna.

 

Really grateful for your help.

[Mark Robinson]

21 minutes ago, The Computer Audiophile said:

Are you using the latest VAD?

 

https://confluence.merging.com/plugins/servlet/mobile?contentId=15139496#content/view/15139496

Yes, using the latest VAD, but maybe im wrong in assuming that Audirvana creates its own AES67 appliance on the network?  

 

VAD essentially turns Apple's Core Audio into a Ravenna device, allowing any native Mac OS sound app (Audirvana included) to communicate with any Merging hardware device broadcasting to the Ravenna network.

 

However, Core Audio is the middleman you want to avoid at all costs IMO. My tests have confirmed that Core Audio degrades sound quality and should be bypassed during critical recording or playback whenever possible.

 

If Audirvana can communicate as a direct AES67 appliance on the network,  then we can avoid going thru VAD's Core Audio driver.  This is what Im looking for.

 

 

[Mark Robinson]

51 minutes ago, The Computer Audiophile said:

Ah, I see. You’re looking for an ASIO equivalent on macOS. 

That's right. From what I understand, a direct AES67 device (with no core audio involved) uses a pure integer transmission, which means no floating point container.  Core Audio puts everything into FP and thru its AU module layers (SRC, summation, wordlenth, and gain) before handing off to Ravenna. No bueno.

[Mark Robinson]

6 minutes ago, Jud said:

Does Audirvana Direct Mode work with Ravenna?

Nope, unfortunately Integer Mode is blocked out with the Merging Hapi (and likely Anubis as well), which is why I was hoping Audirvana could speak directly to the Ravenna network without needing CA.

[Mark Robinson]

2 hours ago, MemoryPlayer said:

P.S.: HQPlayer sounds much better than Origin and Studio!

This is only true if you love the sound of oversampling. As a grammy winning mastering engineer, I can tell you point blank that oversampling pulls the master away from the original analog sound.  

 

How do I know this?  Because I have mastered countless songs from half inch 2 track tape (ATR-102), and the moment you start oversampling the original PCM transfer, you leave the analog capture behind and move towards a DSP sound that many people love because it makes them think their systems sound better than they really are.

 

When I listen to Audirvana VS HQPlayer using my 96khz PCM transfer (via Lavry Savitr ADC), I cannot say HQPlayer sounds better. 

[Mark Robinson]

4 hours ago, The Computer Audiophile said:

Which non-oversampling DAC do you use?

In mastering we use two DACs, a transfer DAC and a monitor DAC. The transfer DAC gets baked into the sound of the final master (that the consumer hears) and the monitor DAC is used only for the engineer's monitoring path.

 

My monitor DAC is a Lavry Quintessence, which is a very transparent delta sigma DAC, great at revealing whats actually there, and nothing more.

 

My transfer DACs are selected on a per project basis according to the sound quality of the incoming mixes (ie: thin, punchy, dark, jumpy, brittle, narrow, congested, etc).  They are the Holo May KTE (R2R NOS), SW1X DAC III Balanced (tube R2R NOS), and Merging DA8P (Delta Sigma OS).

[Mark Robinson]

1 hour ago, Jud said:

 

Some people think oversampling sounds better, some don't. (This is also somewhat dependent on the sample rate output by the ADC.) Some like the sound they get from HQPlayer, some from Audirvana, others don't hear a difference.

 

All of you should let me know when you have convinced everyone else that your way is the right one. 😂

Because Im a dinosaur that still masters from half inch tape, im able to monitor the analog source before any digital touches the signal. Once I start choosing DACs, processors and ADCs, my mastering console allows me to switch between the original tape source and the roundtrip AD DA conversion to analyze the differences.

 

I can tell you all with 100% certainty that the sound of oversampling is emphatically not the sound of the analog source. Does it sound bad?  Obviously not, and in many cases it helps pull the mix apart in a way that could suit the users monitoring path, but to assume it is revealing the source in its truest light is false.

 

If someone prefers the sound of deviation from the source (oversampling), of course that is their right to prefer that deviation as long as they are aware that is one to begin with. Heaven forbid they think they are listening to the truth according to the intention of mix and mastering engineers when in fact they are not. :)

[Mark Robinson]

34 minutes ago, AudioDoctor said:

 

Am I mistaken or doesn't every delta sigma DAC do filtering and sampling on the music? I only know of one that doesn't, and the effort to make it do so was not insignificant.

 

How dare someone enjoy their music, in their house, in a way that they like, that differs from your opinion of how you like to listen to your music, in your house, and in a way you like...

I believe both my DS DACs are filtering and oversampling.  Some projects sound better when I transfer thru them, but most projects I prefer R2R NOS. The R2R NOS transfers are typically a little darker, but this can be easily made up for with a gentle shelving EQ.

 

I have no problem with how people like to enjoy their music in their homes (im not sure how this applies?), but I do have a problem when a delivery format changes the intended sound of a master while operating under the guise of transparency.

 

For example, I have a few thousand masters on Tidal that have been MQA'd after my delivery to the client.  The MQA process represents a clear deviation from the sound I delivered to the artist or label who hired me.  Im allowed to object to what MQA is doing because they are altering my work after the fact. In a similar way, this is what is happening to my masters when they are massively oversampled for playback - the sound deviates from the engineer's intention.

[Mark Robinson]

2 hours ago, Jud said:

 

It's quite interesting that you object to MQA. MQA's filters are so short they in fact do very little filtering, leading to imaging and intermodulation distortion. This distortion is something they have in common with NOS DACs, particularly on RedBook material. However, whether the distortion products are strong enough to be audible is open to question.

 

Oversampling, particularly for RedBook material and if done well, removes much of this distortion and is thus measurably closer to the original. So if NOS is preferred, it's worth asking yourself what (if anything) it is about the sound that makes it so.

 

Of course what I said before about convincing everyone else you're right applies to me too, so if you're not convinced, that's perfectly fine. Peace. 🙂

 

(And lastly, apologies for continuing the off-topic.)

It is hard to make generalizations about NOS vs OS, especially since both have a variety of different implementations that can affect their sound quality.  Ive mostly used delta sigma OS DACs in my work and they are definitely more linear sounding than R2R NOS DACs.  Thats a safe generalization.

 

However, once I started using NOS ladder DACs a few years ago, I noticed that the midrange presentation had a firmer, weightier quality compared to DS.  Coming from 20+ years of cutting tape (still own Stephens 821b, Studer 827mk2, and ATR-102 machines) this midrange weight has alluded me with PCM transfers. When I first heard 1 bit DSD, I got the mids back from tape. When I first heard R2R, I got most of the mids back from tape, but I lost some top end resolution due to some low level hash.

 

In mastering the mids are where the money is. The highs and lows can be more easily contoured with EQ, but there is no fixing the midrange quality, particularly the upper mids.  Using tubes and transformers can help smear the mids back into an analog representation, but now you get distortion, loss of separation and other masking effects.

 

Bottomline for me: I use OS DS for monitoring because of its linearity, low distortion and speed, while accepting the dreaded delta sigma etching thru the mids. I use R2R for a lot of transfers in order to restore the mids back into place, and also as cure for digititus resulting from way too much DSP being used in the DAW these days. Excessive plugin use at 44.1khz (aliasing) is the culprit behind that vast majority of modern-day sound quality issues, but thats another topic altogether.

 

Ok thats enough OT. Thanks for letting me share these thoughts. I use Audirvana Studio daily to check my work on various playback systems.  It lets me listen to my original 2496 prints against the Qobuz/Tidal deliveries on a level playing field.

[Mark Robinson]

8 hours ago, MemoryPlayer said:

Only a doubt, are they sound the same, or Qobuz/Tidal change anything?

I submit both 1644 and 2496 masters to my clients. I would say 1/4 of them pay extra to upload the 2496 separately from the redbook WAVs.  Qobuz doesnt mess with them.  Tidal uses MQA which changes the masters every single time.