Streaming vs dedicated CD player

[imprezap2]

In my system playing a CD in my dCS Rossini gives the best result, in order of performance:
- CD in dCS Rossini
- Ripped files and high res downloads on SSD drive, connected to dCS Rossini
- Ripped files and high res downloads on Aurender N10 (Internal storage)
- Streaming (Qobuz/Tidal)with dCS Rossini
- Streaming with Aurender N10 (very close to Rossini, more a matter of taste, Aurender conductor App is very user friendly and reliable)
- Playing CD/SACD/Blu Ray with Oppo BDP 103D (Audiopraise digital board), Rossini as DAC.

In the last few years streaming has come closer to CD, changing the power supplies for router etc, and the addition of an audiophile network switch helped here.

Usually using the Aurender N10 to stream Qobuz, performance is still very good and so "comfortable", when I come across music I really like I will order the CD.

 

[mdp632]

It's almost 2021 and physical CD playback still is the one to to beat. Quality over convenience every time.

Subjective yes and we can each prefer our own configuration. However, we can agree on the following facts.

1.) Advances in playback equipment for physical compact disc have taken compact disc performance to a higher level. (Esoteric FPGA , Luxman-D10Xu....)

2.) It takes an extreme amount of patience and deep pockets to remove distortion from computer/network audio.

3.) The simplicity of your CD player running a pair of analog outs in your pre-amp. I wonder if the shortest signal path has something to do with it? Network audio , not so much.

4.) Perhaps the issue was never with 16/44 Redbook CD format. The issue was with the playback equipment.

Hardware technology in recent years has breathed new life in the compact disc.

 

[Randy Myers]

But that CPU is inherently noisy. That noise gets into the audio signal. I wish it wasn’t so because the UI is so damn good

Ummm... 5% usage does not make a CPU noisy (that's like driving your car at 10 MPH, your engine is not working at all)... 30% use does not either in reality.... maybe the low power CPU's being used in dedicated built "servers" might get noisy but certainly not in well made appropriate powered machines. My 8 core CPU does not cause the machine to get noisy at all, even when HQPlayer pushes it harder than any other software available... much more so than Roon stand alone. Roon is actually very lite weight in CPU usage compared to most any other music playing software. Even JRivers is more resource demanding.

 

[Randy Myers]

It's almost 2021 and physical CD playback still is the one to to beat. Quality over convenience every time.

Subjective yes and we can each prefer our own configuration. However, we can agree on the following facts.

1.) Advances in playback equipment for physical compact disc have taken compact disc performance to a higher level. (Esoteric FPGA , Luxman-D10Xu....)

2.) It takes an extreme amount of patience and deep pockets to remove distortion from computer/network audio.

3.) The simplicity of your CD player running a pair of analog outs in your pre-amp. I wonder if the shortest signal path has something to do with it? Network audio , not so much.

4.) Perhaps the issue was never with 16/44 Redbook CD format. The issue was with the playback equipment.

Hardware technology in recent years has breathed new life in the compact disc.

Playing from files stored locally versus stream across a network is exactly the same as playing from a CD except you do not have the noise of a physical disk spinning. It is the network and streaming element that starts causing issues, not the digital format being sent to your DAC.

 

[W9TR]

Theoretically a ripped file could sound better than playing physical media since you can error correct over several encoding passes to get a perfect rip. When you spin a disc, you are counting on the disc player’s internal error detection and correction circuitry to do its thing perfectly. I’ve done some listening experiments on my system and have found no difference between the two. Perhaps someone with a more resolving system will have a different outcome.

 

[Randy Myers]

Theoretically a ripped file could sound better than playing physical media since you can error correct over several encoding passes to get a perfect rip. When you spin a disc, you are counting on the disc player’s internal error detection and correction circuitry to do its thing perfectly. I’ve done some listening experiments on my system and have found no difference between the two. Perhaps someone with a more resolving system will have a different outcome.

I certainly would not say that my system is more resolving than yours .... I do notice a difference playing the ripped files versus the disk, and even more so comparing the SACD. I attribute these to a few things. First spinning the physical disk can add noise. Second the DAC being used. The DAC in my Oppo is solid, a ESS 9018, but it is not close to the T+A. Also being able to run the signal through HQPlayer and then on to the DAC at DSD512 is a huge improvement.

With SACD's they are playing at DSD64. While DSD64 is very good it also inherently has an issue with high frequency noise that does fall within the audible range. This is the reason that T+A turns off their WIDE function if DSD64 is detected for fear of damaging your speakers. By ripping the files I get the convenience of playing from the server but more important I can run them through HQPlayer and they end up sounding a whole lot better. I have definitely noticed the better the file going in the better it comes out of HQPlayer! So comparing a SACD rip to a DSD256 or DSD512 download running through HQPlayer simply sounds better.

 

[Mike]

I certainly would not say that my system is more resolving than yours .... I do notice a difference playing the ripped files versus the disk, and even more so comparing the SACD. I attribute these to a few things. First spinning the physical disk can add noise. Second the DAC being used. The DAC in my Oppo is solid, a ESS 9018, but it is not close to the T+A. Also being able to run the signal through HQPlayer and then on to the DAC at DSD512 is a huge improvement.

With SACD's they are playing at DSD64. While DSD64 is very good it also inherently has an issue with high frequency noise that does fall within the audible range. This is the reason that T+A turns off their WIDE function if DSD64 is detected for fear of damaging your speakers. By ripping the files I get the convenience of playing from the server but more important I can run them through HQPlayer and they end up sounding a whole lot better. I have definitely noticed the better the file going in the better it comes out of HQPlayer! So comparing a SACD rip to a DSD256 or DSD512 download running through HQPlayer simply sounds better.

Randy, to be fair, you have to use the same DAC section. Try a digital out (digital RCA) out of your Oppo into your DAC. Then compare.


Sent from my iPhone using Tapatalk Pro

 

[aKnyght]

The reasons you need such power is because Roon is an absolute pig for resources. Like I’ve said, I love the interface, but sonically and it’s resourcing requirements, are not ideal...

...There’s an old post - 18 months ago maybe, where we compared Nucleus vs Innuos Vs Aurender. Aurender won. We had six guys listening.

But that CPU is inherently noisy. That noise gets into the audio signal. I wish it wasn’t so because the [roon] UI is so damn good, but we can hear it when we do comparisons.

> there is also this thread which discusses this same topic of roon software and comparisons to others.
​

respectfully offered for consideration...

these are not objective tests of playback software. they are tests of hardware/software combinations ...and, imo, the differences heard are overwhelmingly if not entirely attributable to the hardware.

i would certainly hope an aurender would outperform a nucleus. aurenders and the like are highly-engineered, purpose-built music servers/streamers -- they are high-end audiophile components with the minimization of electrical noise as an overriding goal. the roon nucleus is an off-the-shelf intel nuc motherboard put in a fanless chassis -- as openly noted by roon, nothing audiophile about this at all.

admittedly, i am a huge fan of roon software; have never had any of the issues noted here and elsewhere, particularly wrt/ resource utilization; and don't see any remotely competative alternative for those wishing to use a (dedicated) computer in their digital stack -- also, roon ROCK has been a game changer.

that being said, no-one should expect high-end audiophile performance from a nucleus. it is a low-end, consumer-grade computer which is under spec'd for many of the use cases owners ask it to perform.

so, to evaluate the SQ of playback software where one software gets to use an aurender and the other software has to use a nucleus is just not a fair fight.

as always, i leave open the possibility that YMMV

images of an n10 and a nucleus...

 

[Randy Myers]

Randy, to be fair, you have to use the same DAC section. Try a digital out (digital RCA) out of your Oppo into your DAC. Then compare.

I have, I used a WireWord Coax cable also so not just a plain Jane RCA cable..... the rip played through Roon and HQPlayer is quite a bit better than spun on the player. Several members of our audio club have done the same comparison and all agree that the rip is better. I am one of the only one that uses HQPlayer though.

Again I do not have super high end spinner although most people would agree I do not have a cheap-o player either. I have done this test using both a Marantz SD-8005 and my current Oppo.

 

[CPP]

I have, I used a WireWord Coax cable also so not just a plain Jane RCA cable..... the rip played through Roon and HQPlayer is quite a bit better than spun on the player. Several members of our audio club have done the same comparison and all agree that the rip is better. I am one of the only one that uses HQPlayer though.

Again I do not have super high end spinner although most people would agree I do not have a cheap-o player either. I have done this test using both a Marantz SD-8005 and my current Oppo.

I have 3 CD players and its interesting the sound I can get by just running S/PDIF or AES ( if equipped) to either of my dacs with NO computer involved at all. Amazing what a different dac can do to the music of a CD player.

 

[Alkyogre]

Hehe, Roon is a lite weight in resource usage verses HQPlayer . HQPlayer is the software that drives the resource requirements over the top .

With just Roon running my CPU use is in the single digits but HQPlayer pushes all 8 CPU's in the 30-40% range, if I am recalling correct (I compared these a few months ago).

What is the point of using all the CPU's? I thought that a light computer with just a USB 2 output is enough to produce 24 bit 192 kHz and will give a very good sound. Even a USB 3 output is overkill. Why do you need all that calculation power?

 

[rbbert]

...Also being able to run the signal through HQPlayer and then on to the DAC at DSD512 is a huge improvement...

Just curious how this might differ from what the PSAudio DirectStream does, without the load on your server's CPU

...With SACD's they are playing at DSD64. While DSD64 is very good it also inherently has an issue with high frequency noise that does fall within the audible range.

Actually the noise is in the ultrasonic range, but still in a frequency range reproducible by most current tweeters. Whether or not that is a problem has been debated since at least the introduction of the SACD (and was probably considered by Sony's engineers during its development), without a conclusion AFAIK.

 

[rbbert]

Theoretically a ripped file could sound better than playing physical media since you can error correct over several encoding passes to get a perfect rip. When you spin a disc, you are counting on the disc player’s internal error detection and correction circuitry to do its thing perfectly. I’ve done some listening experiments on my system and have found no difference between the two. Perhaps someone with a more resolving system will have a different outcome.

Your experience may differ, but I find it very unusual for a clean disc to NOT rip perfectly after one high-speed pass.

 

[the professor]

Just curious - compared to what? I mean - how do you know? I've ripped disks twice and then bit by bit compared files and yes they were 99.9% the same but not the elusive 100%. I never could hear any difference but the files were NOT the same. I gave up that ... wild goose chase of scientific perfection and went with my ears.

I'm truly just trying to learn. All of this makes me want to explore more.

Your experience may differ, but I find it very unusual for a clean disc to NOT rip perfectly after one high-speed pass.

 

[Randy Myers]

What is the point of using all the CPU's? I thought that a light computer with just a USB 2 output is enough to produce 24 bit 192 kHz and will give a very good sound. Even a USB 3 output is overkill. Why do you need all that calculation power?

Enough... maybe, but not enough if you want it done well. Also, DSD is my favorite digital format and what most record companies use for archiving. It all depends on processing... HQPlayer uses all eight cores because it does a ton of processing. USB version has nothing to do with process power. USB 2.0 is one of the popular transport interfaces.

 

[Randy Myers]

Just curious how this might differ from what the PSAudio DirectStream does, without the load on your server's CPU

Actually the noise is in the ultrasonic range, but still in a frequency range reproducible by most current tweeters. Whether or not that is a problem has been debated since at least the introduction of the SACD (and was probably considered by Sony's engineers during its development), without a conclusion AFAIK.

It is considerably different. I have actually talked to Ted Smith about this. Using HQPlayer processes the files in the server and can send the signal in a higher format to the DAC. I send all music to the DAC in DSD512. PS Audio can only accept files up to DSD128. It converts and processes all files to a much higher DSD resolution internally and then converts the signal back down to DSD128 to be decoded by their DAC chips.

The information I related was straight from T+A as they described why their DACs turn off the WIDE function when detecting a DSD64 signal.

 

[rbbert]

Just curious - compared to what? I mean - how do you know? I've ripped disks twice and then bit by bit compared files and yes they were 99.9% the same but not the elusive 100%. I never could hear any difference but the files were NOT the same. I gave up that ... wild goose chase of scientific perfection and went with my ears.

I'm truly just trying to learn. All of this makes me want to explore more.

AccurateRip

 

[rbbert]

Where did you find the (mis)information that most record companies use DSD for archiving? Of course some albums are archived that way, but most are not (Sony does archive DSD copies of all their analog masters)?

 

[rbbert]

The information I related was straight from T+A as they described why their DACs turn off the WIDE function when detecting a DSD64 signal.

I haven’t read what T+A says, but I doubt they would make the elementary mistake of saying the extra noise from DSD is in the audible range (the noise shaping was specifically designed to put it in the ultrasonic range, and it is). As I posted, though, most (all?) tweeters do respond well into the ultrasonic range. Whether or how much the DSD noise will affect the sound is very much open to debate, although T+A has clearly stated their opinion. Note that every DSD sampling rate will have noise shaping, the only difference is at which frequencies the extra noise appears.

 

[Randy Myers]

I haven’t read what T+A says, but I doubt they would make the elementary mistake of saying the extra noise from DSD is in the audible range (the noise shaping was specifically designed to put it in the ultrasonic range, and it is). As I posted, though, most (all?) tweeters do respond well into the ultrasonic range. Whether or how much the DSD noise will affect the sound is very much open to debate, although T+A has clearly stated their opinion. Note that every DSD sampling rate will have noise shaping, the only difference is at which frequencies the extra noise appears.

I certain might be mis-remembering them saying audible range instead of range that can affect the audible music and/or damage speakers. But the fact remains that they turn off the WIDE function when they detect DSD64, and consider DSD128 or more so DSD256 the minimal level for proper DSD playback. To take advantage of what DSD is all about. And why the heck are you splitting hairs here; it seems like you are trying to make an issue where one does not exist. The bottom line is their view, and mine is that DSD128 is the minimal, and DSD256 or DSD512 is where the amazing benefits of DSD are shown, and why HQPlayer with incredible DACs such as the T+A (among other top flight DACs) are very popular.

From T+A Manual:
"By its nature the DSD format involves a noise floor which rises above
the range of human hearing as frequency rises. Although this noise
floor is not directly audible, it does subject the treble units in the
loudspeakers to a significant load. It is also possible for the high
frequency noise to cause distortion in many low-bandwidth amplifiers.
The lower the DSD sampling rate, the more severe the inherent noise,
and it cannot be disregarded, especially with the 2,8 MHz (DSD64)
format - as used on the SACD. As the DSD sampling rate rises, the
high-frequency noise becomes increasingly insignificant, and with 11,2
MHz (DSD256) it is virtually irrelevant. In the past it has been standard
practice to apply digital and analogue filtering processes in an attempt
to reduce DSD noise, but such solutions are never entirely without side
effects on sound quality. For the DAC 8 DSD we have developed two
special techniques designed to eliminate the sonic disadvantages:
1. The  True-DSD technique, consisting of a direct digital signal
path without filtering and noise-shaping, plus our True 1-bit DSD
D/A converter
2. Analogue reconstruction filter with adjustable bandwidth.


The bandwidth cannot be switched to the 'WIDE' mode when
playing DSD files with 2,8 MHz (DSD64). (see section entitled
‘D/A converter settings’)


With thebutton, the bandwidth of the analogue output filter can
be switched between 60 kHz (‘CLEAN’ mode) or 120 kHz ('WIDE'
mode).
The ‘WIDE’ setting allows a more spacious music reproduction.

DSD signals are characterised by a high-frequency noise floor in the
ultra-sound region. This noise is inherent in the DSD principle, and is
present in the recording itself. In basic terms the higher the sampling
rate, the lower the high-frequency noise. At 2,8 MHz (DSD64) this
noise is by no means negligible, and may constitute a problem for
amplifiers and loudspeakers connected to the system. For this reason,
the bandwidth cannot be switched to the 'WIDE' mode when playing
2,8 MHz DSD files (DSD64)."