JRiver owners: do you ever use DSP effects?

o0OBillO0o said:
I've followed your comments on other forums. Your software of offline conversion is certainly intriguing.
Click to expand...

Thank you, Bill, for attention to my converter. I deep involved in development process and try do my work best, as I can.

o0OBillO0o said:
I've followed your comments on other forums. Your software of offline conversion is certainly intriguing.

From what I understand, by converting the music file to the format that is either best or at the maximum rate for a given DAC, the software allows a resource limited DAC to run in a less constrained mode. The is as a result of using even more precise calculations, using a CPU, through the software.

This is all under the concept of how a DAC handles data. Most DACs automatically, by design, convert the audio files to very high bit depths and sample rates inside the DAC. By "priming" the DAC in advance, with a pre processed audio file that has been calculated with greater precision that what the DAC could achieve on it's own resource limited capabilities, we may experience an increase in DAC performance.
Click to expand...

Yes.

I’m not hardware developer. However many time close worked with DSP hardware developers.
Permanently I collect information by colleagues by industry, who develop hardware, and by necessity learn data sheets on different components.

As me known, real precision used in consumer audio hardware up to 32 bit integer.

DSP processors allow work with 32-bit float (possibly some can process even 64-bit float).
For FPGA more troubles (or even impossibly) release floating point math.

Of course, I can’t guarantee it for all devices.
If I mistaken in written above, will glad open for me new details.

Using floating point formats allow avoid overloading.
Overloading can appear during oversampling. Appear new samples with level above maximal of source sample.

Also overload can be during filtration in integer mode: multiply, summation long IIR filters.

Floating point value has level range -1.0 … 1.0. If will overload, it will 1.001 as example. Thus we control overload and can correct it different ways.

64-bit float have more limit for overloading and more suitable for DSD256 and 512.

DSD128 has abilities like 32-bit float. I checked it.
DSD256 and 512 not checked due absent of properly measurement tools.

Main advantages of PC processing is almost unlimited memory with transparent for programmer access and power multi-core CPUs.

Sometime I get improving of performance, however resampling algorithms, especially DSD's, insatiable :)

o0OBillO0o said:
No ruling on increase of sound quality. but as you say, try your software before you buy
Click to expand...

Upsampling don’t improve sound quality converted file as itself. However offline resampling and filtration allow put signal straight to analog filter of DAC.

DAC structure must allow do it, of course. Get real detail structure often is problem.

Other way measure system from player to output DAC. But it is possibly only with enough expensive measurement hardware.

Remains listening tests only. Also not all what measured, possible listened :)
 
o0OBillO0o said:
Side question- if I have a 32bit capable DAC (PCM1795 | Audio DAC | Audio Converter | Description & parametrics) , how would I use AuI to best serve this DAC?
Click to expand...

1. I'm not sure, but me seems (by scheme and description) 8x internal resampler able turn off for using external filter chip (DF1704, as example).
How it applied in your DAC-device (not chip) unknown for me. May be DAC-device manual contains some information about ability pass by 1795's 8x filter.

However I don't know what analog low frequency filter better in your DAC-device.
Here, me seems, better compare all sample rates. Not fact that maximal sample rate will have better sound.

Also, if DAC-device use 1795's DSD capability, it makes sense compare DSD vs. PCM in different modes.


2. Bit depth, I suppose, better compare by hear between 24 and 32 bit. Comparing better perform on quiet places. Level volume possible increase for better listening.

Theoretically 24 bit enought for 123 dB dynamic range of 1795's. However 32 bit possibly give some advantages for quiet places.
 
YuriKorzunov said:
1. I'm not sure, but me seems (by scheme and description) 8x internal resampler able turn off for using external filter chip (DF1704, as example).
How it applied in your DAC-device (not chip) unknown for me. May be DAC-device manual contains some information about ability pass by 1795's 8x filter.

However I don't know what analog low frequency filter better in your DAC-device.
Here, me seems, better compare all sample rates. Not fact that maximal sample rate will have better sound.

Also, if DAC-device use 1795's DSD capability, it makes sense compare DSD vs. PCM in different modes.


2. Bit depth, I suppose, better compare by hear between 24 and 32 bit. Comparing better perform on quiet places. Level volume possible increase for better listening.

Theoretically 24 bit enought for 123 dB dynamic range of 1795's. However 32 bit possibly give some advantages for quiet places.
Click to expand...

Yuri- unfortunately, I contacted the manufacturer, Ron Cornelius at McIntosh Labs, and he related that their implementations of the PCM1795 (C50&C2500) do not allow to bypass the 8X internal resampler and do not take advantage of the DSD capability. There is no plan to update the products through firmware updates and he recommends either using JRiver to convert on the fly. (Or us AuI to convert offline. ;) ) I am also certain the solution to buy the McIntosh D150 (DXD & 2XDSD DAC) is possible. The so what is, "Most people are fixated on a number but in reality they cannot hear or tell the difference (w/ repeatability)."

Bit depth: I am not convinced that I could hear the difference in audio when we are talking even 100dB below the maximum level. Who can? Which leads to the concept that, yes, a 24bit DAC has 123dB of signal to noise, but the analog circuit has a much lower signal to noise ratio and becomes the limiting factor in the electronic component chain. (Of note: Rooms and speakers introduce greater distortion.)

Thank you for answering my questions, is there anything I can do to my Audio files for my DAC Using AuI, given the addressed limitations?



Back to your AuI converter, "Sometimes I get (a performance gain with AuI)*, however resampling algorithms, especially DSD's, (are)* insatiable"

Could you explain more, how are some of the resampling algorithms impossible to satisfy?



*I understand English is not your primary language, but is certainly better than my Russian. I made some edits that I felt made your point clearer. ;)
 
o0OBillO0o said:
Bit depth: I am not convinced that I could hear the difference in audio when we are talking even 100dB below the maximum level. Who can? Which leads to the concept that, yes, a 24bit DAC has 123dB of signal to noise, but the analog circuit has a much lower signal to noise ratio and becomes the limiting factor in the electronic component chain. (Of note: Rooms and speakers introduce greater distortion.)
Click to expand...

32 bit close DSD128. I’m also not sure: possibly hear difference or not.

o0OBillO0o said:
Thank you for answering my questions, is there anything I can do to my Audio files for my DAC Using AuI, given the addressed limitations?
Click to expand...

I suppose only seamless albums resampling (True Gapless Conversion) only :)

Also can try minimal phase filters. Last time I see rising of interest to this kind of filters. However while a few feedback.


o0OBillO0o said:
Back to your AuI converter, "Sometimes I get (a performance gain with AuI)*, however resampling algorithms, especially DSD's, (are)* insatiable"

Could you explain more, how are some of the resampling algorithms impossible to satisfy?
*I understand English is not your primary language, but is certainly better than my Russian. I made some edits that I felt made your point clearer.
Click to expand...
No problem.
In Russia we learned English from 10-11 years old (now from 8-9), but without intensive communications it’s useless :)
I always appreciated for pointing to my language mistakes. Now impossibly without English.

DSD algorithms consume huge computing resources due higher (than PCM) sample rates.
Here: more sample rate - more CPU resources.
 
You must log in or register to reply here.
Back
Top