I’ve started using Audirvana (I’ve used Amarra in the past as well as iTunes+BitPerfect). Sound is clearly superior with better separation, perceived detail, etc. vs anything else I’ve listened to in computer audio. The sound is organic but sadly it doesn’t address the inherent harshness of digital, I would say it probably makes it worst.
Are there any settings (or filters-Processing) that can help address this issue? (I’m using it with an Arcam Burr Brown S/D DAC as well as a R2R NOS TDA1543 via tube amplification and results are similar), thank you!
Thanks for your answer. It’s on my considerations: my DAC does not support DSD but I’m thinking of buying one for that purpose, did you see a difference? Also why are you suggesting r8Brain since Audirvana can perform the upsampling on the fly?
You might look into an EQ plugin that can also impart some additional warmth via tube/transformer emulation… Like the Pulsar Massive which is an emulation of the Manley Massive Passive EQ for instance… Do you have any room correction or acoustic treatments? Have you thought of doing any small signal tube rolling in your kit? You might also try minimum phase with a slow rolloff setting with the r8brain resampler…
Thanks, this is really useful: the r8brain resampler is very interesting and seems to be working on that direction and I’ve downloaded the Pulsar Massive.
How are the digital files ending up at the DAC? If raw USB from a computer, harshness is often due to noise from USB transmission coupling from source to load via the /USB shell on the connector.
Caused usually by a difference in chassis potentials between the DAC and source.
There are many ways around this, USB signal isolators, reclocker, regenerator type devices from SOTM, Uptone, ifi, Intona, JCAT and so on and on!
(From the 2014 Stanford University CCRMA (Center for Computer Research in Music and Acoustics) paper linked below)
While analog audio produces a constantly varying voltage or current, digital audio produces a non-continuous list of numbers. The maximum size of the numbers will determine the dynamic range of the system, since the smallest signal possible will result from the lowest order bit (LSB or least significant bit) changing from 0 to 1. The D/A converter will decode this change as a small voltage shift, which will be the smallest change the system can produce. The difference between this voltage and the voltage encoded by the largest number possible (all bits 1’s) will become the dynamic range.
This leads to one of the major differences between analog and digital audio: as the signal level increases, an analog system tends to produce more distortion as overload is approached. A digital system will introduce no distortion until its dynamic range is exceeded, at which point it produces prodigious distortion. As the signal becomes smaller, an analog system produces less distortion until the noise floor begins to mask the signal, at which point the signal-to-noise ratio is low, but harmonic distortion of the signal does not increase. With low amplitude signals, a digital system produces increasing distortion because there are insufficient bits available to accurately measure the small signal changes.
There is a difference in the type of interference at low signal levels between analog and digital audio systems. Analog systems suffer from thermal noise generated by electronic circuitry. This noise is white noise: that is, it has equal power at every frequency. It is the “hiss” like a constant ocean roar with which we are so familiar. The low-level noise generated by digital systems is different: it is correlated with the signal because it is really a form of signal distortion produced by quantizing errors and has a distinctly unpleasant sound: something like a “gritty” or “grainy” sound. Most listeners find digital noise to be less acceptable than a comparable amount of analog white noise. Do we need to use noise reduction with digital systems as we found necessary on analog recorders? Well, the answer is yes, sort of. As we will later see, there are digital techniques that may be used to reduce the undesirable low-level distortions without processing the signal as we did for analog noise reduction. https://ccrma.stanford.edu/~jay/subpages/Lectures/Lecture8-Digital_audio_systems.pdf
My question was really rhetorical, I don’t hear no harshness, and the OP asked how to get rid of digital harshness, surely the question should have been " how to remove the harshness from my system “? Because it appears there is a general perceived harshness in the OPs system. Being as pretty much all audio today is digital at some stage of it’s journey to your ears, the producer of said audio would not allow it out into the world if it sounds 'Harsh” ( except maybe Aphex Twin!)
I think I understood his sensibility… I personally prefer the sound of PDM (1-bit DSD) over LPCM, because PCM generally sounds less organic (at any sample-rate) on my system that can be either PCM-centric or DSD-centric… I tune my system for DSD playback… I am so glad that I can modulate all of my LPCM material to DSD128 with r8Brain in Audirvana.
Digital-Audio has evolved out of convenience for the most part… There are trade-offs in both arenas… It’s getting better all the time, Audirvana is a prime example of the evolution of digital-audio file playback engines, and modern high-resolution DAC platform designs have evolved to provide very transparent and focused D/A conversion of the originally encoded signals stored on digital media… It’s my opinion that a pure DSDxxx encoded signal reproduced by a pure DSDxxx signal path of a pure DSD DAC platform, brings us closer to the analog signal… Some might say that a PGGB up-sample of PCM file, sounds as good or better… It’s still PCM and it has an inherent character due to the encoding process.
Yes… And how these transducers are designed and managed in application will impart ‘color’ to say the least… If we were to be able to create a perfect translation of the encoded digital signal to the analog space that we audition from these transducers, there remains the fundamental encoding process that imparts qualitative fingerprinting, as in the difference between PDM encoding and LPCM encoding… These fundamental encoding processes are quite different in sonic signature, which is intrinsically tied to the technology used in the translation to an analog signal through the output circuitry and ultimately the playback architecture/transducer application environment.
Tube emulation is a color that may suit the conditioned ears of some… As is the application of speaker and headphone management in the control of acoustical environment behaviors, etc…
For those of you who find the pulsar massive a little too expensive …
I use the free Warmy EP1A Tube EQ V2 plugin
Even with the default settings the „harshness“ is gone. At least in my subjective opinion.
Maybe just a placebo but it works for me.
