Regular dropout on DSD files

[Agoldnear]

… @bitracer

Maybe you are right… perhaps there is a macOS problem addressing the unified memory caches… However, this is highly doubtful, due to the vertical nature of the hardware and macOS interfacing… I’m not saying the macOS is bug-free… Having employed Apple computer platforms for as long as I have (since the Apple IIe and Mac Plus), I’m quite familiar with bugs cropping up after OS updates and versions upgrades and hardware advancements at the component level, etc…

I am presuming the Core Audio and/or MusicPlayer API’s are fundamentally unchanged on the application (Audirvana) side of the macOs interface, the only difference would be in how the Core Audio and/or MusicPlayer API’s interface to the hardware interface of the M1 SoC unified memory cache… This seems to be operationally fine, as @Dandaman has no problem with DXDxxx file playback…

In the digital-audio realm, available RAM for flawless playback of any high-resolution audio file (PCMxxx, DXDxxx and DSDxxx) has always been a factor, outside of definite operational bugs and system configuration bugs that precipitate drop-outs and loss of sync problems… The massive DSD file(s) memory requirement, is the primary reason realtime DSP processing of DSDxxx is nonexistent and is relegated to basic editing… the file size and 1-bit Fs format prohibits the employment of realtime DSP (at this point in audio history…) …When I can employ my 112dB’s Redline Monitor AU plugin to create a virtualized binaural HTRF playback on a stereo DSD64 file, with an 8GB M1 platform or any M series platform with even greater memory capacity like Mac Studio or Mac Pro, then I will believe you are on the right track, (regarding a potential system bug…)

From what I can extract from searching audio-centric forums and Apple Community posts, is M1 memory related issues are coming from 8GB M1 platform users, not 16GB M1 platform users.

I am deploying 11.4GB of buffer… the primary reason for this amount of dedicated RAM, is to enjoy flawless DSDxxx playback and remodulation of PCMxxx, DXD 352.8 to DSD128 using SoX in Audirvana…

[Phil83]

Bonjour,

Pardonnez-moi d’intervenir .
Je viens de lire ce post et j’ai un soucis un peu similaire sous Audirvana Studio avec Windows 10 installé sur un serveur dédié.
En fait mon soucis se présente uniquement à la lecture de fichiers DSD256.
J’avais contacté le support technique et nous avons échangé une trentaine de mails, en faisant des essais, des tests mémoires etc…
Cela n’a toujours pas été résolu à ce jour.
Cela fonctionnait bien avant la mise à jour 1.12 ou 1.13 (je ne sais plus exactement).
J’ai même tenté l’essai sur Roon et le problème ne s’y présente pas. Mais j’aime le son que produit Audirvana Studio sur mon système, d’où ma raison de rester.
Je pense donc, pour moi, que le problème viens d’Audirvana Studio lui-même, et espère une mise à jour en profondeur afin de résorber définitivement ces désagréments.

Bien musicalement.

Philippe.

[Agoldnear]

… @Phil83

How much pre-load buffer are you deploying?
What size is your available system RAM?

My concern regarding these side-cases, is that Audirvana Studio quality-of-sound will be compromised in trying to accommodate a niche DSD256 catalog… Apparently, you can perceive a difference in sound-quality in an application like ROON that is trying to do everything for everyone… Do we want to compromise the sound-quality of Audirvana just to play a miniscule catalog of DSD256 files?

If in fact there is not enough buffering for DoP 1.1. processing of DSD256, where the file size of a raw .dsf DSD256 file is 161.50MB/min, before interleaving into the PCM file carrier (where a single 7 minute raw .dsf DSD256 track would be 161.5MB x 7 (min) = 1.13GB) [11.288MHz DSD256 1-bit Fs requires a 705.6kHz PCM signal for DoP 1.1]… I would expect when there is an extensive enough DSD256 catalog to make rational sense to make changes to Audirvana Studio so to accomodate these huge files, without compromising the established level of sound-quality that Audrivana now presents, we will see this happen.

Another thing to consider is ones ability to actually perceive harmonic and dynamic subtleties delivered by a DSD256 recording versus the same recording delivered at DSD128 or visa-versa… My personal take, is, this is getting into the realm of very subjective interpretation and intrinsically tied to system component configurations playback environment, etc and at the production level and the decisions made in the production processes of creating the DSDxxx product.

[Phil83]

Bonjour, merci beaucoup pour votre retour.

Mon système possède 12Go de Ram et la mémoire tampon est réglée sur 4Go.
Je fait la lecture directement via le driver Asio (Dac Teac (NT-503), en natif, sans passé par une conversio DoP. L’objectif étant de lire le format audio pleinement.

Musicalement.

Philippe.

[Agoldnear]

… @Phil83

DoP (DSD over PCM) is a means for transferring the large DSDxxx format files effectively using a common digital-audio format (PCM)… The raw .dsf DSDxxx is packaged/interleaved into the PCMxxx signal without conversion to PCM:
https://dsd-guide.com/dop-open-standard#.YtbXVi-B1B0

From the DoP white-paper linked above:

Windows by nature does not fully support USB Audio 2.0 and what it does support is limited to PCM only at a sample rate of 96kHz or less. There is no native driver support for higher resolution PCM and it is clear from the beginning that a custom driver needs to be created for this platform whether it is for regular PCM or DSD…

ASIO is not directly a hardware driver, but sits between the audio playback application and the hardware driver. Each hardware manufacturer still needs to develop a custom hardware driver for their own hardware, but ASIO then creates a common interface standard for all application software.

…Windows platform basically offers a solution with the ASIO driver and the raw data format supported by USB Audio 2.0. Not as ideal as having a dedicated DSD path via USB, but this is safe and straightforward…

Have you tried setting Audirvana Studio to send DSDxxx via DoP 1.1?

The only true raw DSD transfer protocol is “I2S” now typically transferred via HDMI interconnect:

I²S

Generally, I don’t believe you are deploying adequate pre-load buffering for flawless DSDxxx playback with 4GB.

[Phil83]

Bonjour,

Merci pour vos éclaircissements.
J’avais essayé également en DoP et le résultat était le même pour les DSD256.
Peut-être que je ne comprends pas très bien, mais pourquoi cela fonctionnait bien auparavant, pourquoi cela fonctionne bien sous Roon, et pourquoi cela fonctionne aussi sous Teac HR Player?
Quand une conversion DSD est faite, j’ai sur l’afficheur du Dac les lettres DoP. En lecture “pleine” cela ne s’affiche pas, (inscription DSD 11,2Mhz à l’écran par exemple)

C’est quand-même pas sympa que les constructeurs proposent du matériel si ce n’est pas lu pleinement.
J’ignorais également qu’il y avait uniquement I2S qui proposait la lecture pleine.

Musicalement.

Philippe.

[Agoldnear]

…

The DAC extracts the DSDxxx signal from the DoP signal and plays the raw .dsf file that is being extracted from the DoP file, therefore the DAC reports that it is receiving a DoP 1.1 file…

In regard to DSD256 playback in ROON and TEAC HR Player… Most likely these players are accommodating larger buffering, at a cost in subjective fidelity…

Are you playing SACD .iso files or .dsf files extracted from .iso files that have been extracted to DSD256? If so this could be a problem, as SACD .iso files must be extracted at the native recording Fs of 2.822MHz.

Using DoP1.1 try doubling your pre-load buffer to 8GB

Also, if you are employing the “Bulk PET” transfer protocol to get signals to the TEAC DAC, you are adding yet another layer of operational transport software, when using ASIO… “Bulk PET” usurps the USB interface controller to manage packet flow across the USB bus…

[Phil83]

C’est vrai, que d’un point de vue de la qualité sonore, comme je l’ai dit dans le fil de discussion, c’est beaucoup mieux sous Audirvana.
Je lis des fichiers en .dsf
J’essaierai votre méthode en doublant la mémoire tampon en DoP 1.1
Je vous tiendrai informé.

Musicalement

Philippe.

[Phil83]

Bonjour,

Je reviens vers vous pour vous donner des nouvelles.
J’ai donc essayé votre méthode, et sans surprise cela est un échec.
J’ai oublié de mentionner également qu’il y a aussi un soucis avec la remote Audirvana au moment de la lecture de DSD256. En effet les informations ne sont plus remontés (pas de jaquette, de titres etc…)
Je n’ai pas d’autres choix que d’attendre qu’un jour cela refonctionne.

Musicalement.

Philippe.

[Agoldnear]

… @Phil83

I’m not surprised either… I don’t believe that you have enough pre-load memory allocated to play the DSD256 files flawlessly…

Where did you acquire these DSD256 files…? Are these files from a legitimate reseller of DSD product, like NativeDSD ? Do you know if these .dsf files have been extracted from a SACD ISO file to iterate the DSD256 .dsf? … Just because you are playing a .dsf file, does not mean that it is actually the native DSD256 file as recorded and that the metadata is not corrupted by a bad ISO extraction process, especially if you didn’t purchase this file from a legitimate reseller of DSD product.

DSD Downloads

I’m limited in my insights into the Windows OS, so you will need help from PC/Windows users… However, I see there is the Microsoft Windows Audio Session API (WASAPI) and Kernel streaming being used by others rather than ASIO…

[Phil83]

Bonjour Agoldnear,

Mes fichiers proviennent de NativeDSD en effet.
J’ai également essayé avec une autre machine doté de plus de RAM (32Go) et alloué toute la mémoire tampon disponible. Le problème reste présent.

A mon sens et après tous les essais, ce n’est pas un problème lié au matériel, c’est interne à Audirvana Studio!
Seule une correction sur cette partie du logiciel pourra résoudre définitivement ce désagrément, malheureusement.

J’attendrai sagement.

Musicalement.

Philippe.

[Agoldnear]

…

Well… If you believe you have made all of the proper allocations of pre-load memory and system control parameters that can be set in Audirvana on either computer, then I agree that there is some software problem… Could ASIO be the culprit? Or does DSD256 put extreme taxation on the computer platform because of it’s Fs of 11.28MHz and 1-bit file format? Is there rationale for DSD256 playback in Audirvana vs. DSD128? … I’m presuming you do not have issues playing DSD128 (5.64MHz) files?

Have you tried using the TEAC “Bulk PET” transfer API with DSD256?
Are you connecting to the DAC via Ethernet on a network with a router or connecting the DAC directly via USB?

[Phil83]

Effectivement, aucun problème avec les DSD64 et DSD128. Ça fonctionne bien.
Où puis-je trouver cette fonction API Bulk Pet? C’est dans les paramètres ASIO? Il me semble avoir vu ça quelque part.

Musicalement.

Philippe.

[Phil83]

Mon Dac est connecté en USB

[Agoldnear]

…
It does not look like the TEAC Bulk Pet USB transfer technology is supported in the NT-503… I assumed that it was… It is supported in the newer NT-505-X…

The fact that you can play DSD128 with no issue, is revealing… (DSD256 is twice (2x) the data density of DSD128)

Does your music library exist on your Windows system storage device or an external storage device (HDD or SDD)? And if the storage device is external, how are you connecting it (USB, Firewire, SATA)? … The DSD256 playback anomaly may be related to bus throughput and hierarchy of the connected components on the the data transfer bus that the music library storage device and the DAC exist on… In the bus hierarchy, the DAC should have highest priority… If your music library storage device is external to the computer, it should be on the fastest transfer bus available and if shared on a common transfer bus, the library storage device, must be lower in bus hierarchy/priority than the DAC.

[Phil83]

C’est ce que j’ai constaté dans la petite application Asio de Teac. Seul la fonction Isochrone est active et on peut régler le buffer sampler (réglage au maximum)
Pour répondre à vos questions concernent ma bibliothèque musicale, elle se trouve sur mon SSD Système. Je n’utilise pas de périphérique de stockage externe.

Musicalement.

Philippe.

[Agoldnear]

…

It’s never a good idea to store any of your music library files on the same storage device as your OS system, especially high-sample rate DSDxxx and PCMxxx/DXDxxx files… This is something that I preach to everyone on this forum… Keep in mind the data density of DSD256 .dsf files…

Apparently you are speaking about the TEAC ASIO driver for Windows… I am an Apple macOS platform user.…You will need more experienced insights from audiophile PC/Windows platform users than I am able to provide for you.

[Phil83]

Exacte Agoldnear,

Il est prévu de passer sur un SSD Interne toute ma bibliothèque.
Je verrai si à ce moment-là ça change quelque chose.
En tout cas merci d’avoir échangé avec moi pendant quelques temps, c’est vraiment sympa.

Musicalement.

Philippe.

[Antoine]

Hello @Dandaman,

Can you send us a copy of the “Debug Info” when you have this issue? You can get it in Audirvana Settings>My account

Note: By clicking on the Debug Info button, Audirvana automatically copies all its content, all you have to do is paste it in response to this message.

Also, can you send us a mail at support@audirvana.com so we can give you a link to send us out the DSD file you have?

[Dandaman]

I’ve sent the email you requested. Here is debug info:

Audirvana Studio 1.13.3 (11303)

macOS Version 12.4 (Build 21F79) with 8GB physical RAM

Connected account of : Daniel Foster

SIGNAL PROCESSING:

Polarity Inversion:
	Globally: OFF
	Per track: ON
Effects plugins ACTIVE in realtime mode with legacy macOS API
	AU plugin #0: Effect - Accurate Sound HLConvolver ()
	AU plugin #1: None
	AU plugin #2: None
	AU plugin #3: None

UPSAMPLING:
r8brain not in use
r8brain filter parameters
Bandwidth = 99.5%
Stop band attenuation 215.701dB
Phase minimum phase

AUDIO VOLUME:
Max allowed volume: 100
Replay Gain: None
SW volume control: OFF

LIBRARY SETTINGS:
Sync list: 1 folders
AUTO: /Volumes/Music Bank/FLAC Library
iTunes/Music library synchronization: not synchronized
Library database path: /Users/danielfoster/Library/Application Support/Audirvana/AudirvanaDatabase.sqlite

Remote Control server:
Listening on 127.0.0.1 on port 59170

ACTIVE STREAMING SERVICES
Qobuz: Connected as Qobuz Studio

=================== AUDIO DEVICE ========================

CoreAudio audio path, buffer I/O frame size: 512

Max. memory for audio buffers: 3440MB
Exclusive access: Enabled
Integer mode: Enabled

Preferred device:
EXASOUND AUDIO DESIGN EXASOUND
Model UID:exaSound-N1
UID:exaSoundDevice-N1

Active Sample Rate: 44.1kHz
Hog Mode is off

Bridge settings:
Sample rate limitation: none
Sample rate switching latency: 2s
Limit bitdepth to 24bit: OFF
Mute during sample rate change: ON

Selected device:
Local audio device
ID 0x3f exaSound
Manufacturer: exaSound Audio Design
Model Name: exaSound Audio Design exaSound
Model UID: exaSound-N1
UID: exaSoundDevice-N1

10 available sample rates up to 768000Hz
44100
48000
88200
96000
176400
192000
352800
384000
705600
768000
Audio buffer frame size : 14 to 4096 frames
Current I/O buffer frame size : 512
Volume Control
Physical: Yes
Virtual: Yes
MQA capability
Auto-detect MQA devices: No
Not a MQA device, user set to not MQA
DSD capability: DSD via PCM 1.1
Device audio channels
Preferred stereo channels L:1 R:2
Channel bitmap: Ox3, layout:
Channel 0 mapped to 0
Channel 1 mapped to 1

Audio channels in use
Number of channels: 2
Use as stereo device only: No
Simple stereo device: Yes

1 output streams:
Number of active channels: 2, in 1 stream(s)
Channel #0 :Stream 0 channel 0
Channel #1 :Stream 0 channel 1

Stream ID 0x40 2 channels starting at 1
20 virtual formats:
2 ch Mixable linear PCM Interleaved 32 little endian Signed Float 768kHz
8 ch Mixable linear PCM Interleaved 32 little endian Signed Float 768kHz
2 ch Mixable linear PCM Interleaved 32 little endian Signed Float 705.6kHz
8 ch Mixable linear PCM Interleaved 32 little endian Signed Float 705.6kHz
2 ch Mixable linear PCM Interleaved 32 little endian Signed Float 384kHz
8 ch Mixable linear PCM Interleaved 32 little endian Signed Float 384kHz
2 ch Mixable linear PCM Interleaved 32 little endian Signed Float 352.8kHz
8 ch Mixable linear PCM Interleaved 32 little endian Signed Float 352.8kHz
2 ch Mixable linear PCM Interleaved 32 little endian Signed Float 192kHz
8 ch Mixable linear PCM Interleaved 32 little endian Signed Float 192kHz
2 ch Mixable linear PCM Interleaved 32 little endian Signed Float 176.4kHz
8 ch Mixable linear PCM Interleaved 32 little endian Signed Float 176.4kHz
2 ch Mixable linear PCM Interleaved 32 little endian Signed Float 96kHz
8 ch Mixable linear PCM Interleaved 32 little endian Signed Float 96kHz
2 ch Mixable linear PCM Interleaved 32 little endian Signed Float 88.2kHz
8 ch Mixable linear PCM Interleaved 32 little endian Signed Float 88.2kHz
2 ch Mixable linear PCM Interleaved 32 little endian Signed Float 48kHz
8 ch Mixable linear PCM Interleaved 32 little endian Signed Float 48kHz
2 ch Mixable linear PCM Interleaved 32 little endian Signed Float 44.1kHz
8 ch Mixable linear PCM Interleaved 32 little endian Signed Float 44.1kHz

20 physical formats
2 ch Mixable linear PCM Interleaved 32 little endian Signed Integer 768kHz
8 ch Mixable linear PCM Interleaved 32 little endian Signed Integer 768kHz
2 ch Mixable linear PCM Interleaved 32 little endian Signed Integer 705.6kHz
8 ch Mixable linear PCM Interleaved 32 little endian Signed Integer 705.6kHz
2 ch Mixable linear PCM Interleaved 32 little endian Signed Integer 384kHz
8 ch Mixable linear PCM Interleaved 32 little endian Signed Integer 384kHz
2 ch Mixable linear PCM Interleaved 32 little endian Signed Integer 352.8kHz
8 ch Mixable linear PCM Interleaved 32 little endian Signed Integer 352.8kHz
2 ch Mixable linear PCM Interleaved 32 little endian Signed Integer 192kHz
8 ch Mixable linear PCM Interleaved 32 little endian Signed Integer 192kHz
2 ch Mixable linear PCM Interleaved 32 little endian Signed Integer 176.4kHz
8 ch Mixable linear PCM Interleaved 32 little endian Signed Integer 176.4kHz
2 ch Mixable linear PCM Interleaved 32 little endian Signed Integer 96kHz
8 ch Mixable linear PCM Interleaved 32 little endian Signed Integer 96kHz
2 ch Mixable linear PCM Interleaved 32 little endian Signed Integer 88.2kHz
8 ch Mixable linear PCM Interleaved 32 little endian Signed Integer 88.2kHz
2 ch Mixable linear PCM Interleaved 32 little endian Signed Integer 48kHz
8 ch Mixable linear PCM Interleaved 32 little endian Signed Integer 48kHz
2 ch Mixable linear PCM Interleaved 32 little endian Signed Integer 44.1kHz
8 ch Mixable linear PCM Interleaved 32 little endian Signed Integer 44.1kHz

Local devices found : 4
Device #0: ID 0x33 BlackHole 2ch Manufacturer: Existential Audio Inc. Model UID: BlackHole2ch_ModelUID UID: BlackHole2ch_UID Model Name: Existential Audio Inc. BlackHole 2ch
Device #1: ID 0x4c Mac mini Speakers Manufacturer: Apple Inc. Model UID: Speaker UID: BuiltInSpeakerDevice Model Name: Apple Inc. Mac mini Speakers
Device #2: ID 0x6e VB-Cable Manufacturer: VB Audio Model UID: com.vbaudio.vbcable.model:38114B1C-AF02-41E4-A1BC-64B91CF802F6 UID: com.vbaudio.vbcable:B6E6D02E-CD32-4C95-B60B-2FC2621CA4E0 Model Name: VB Audio VB-Cable
Device #3: ID 0x3f exaSound Manufacturer: exaSound Audio Design Model UID: exaSound-N1 UID: exaSoundDevice-N1 Model Name: exaSound Audio Design exaSound

UPnP devices found : 0