Yes, and naim themselves did many a demonstration to show us what they could get out of a CD!
I spoke to (one of) the guy(s) involved in designing Audio Noteās DACs recently.
While I didnāt follow everything he said, his general thrust was that he thought there was a lot of improvement over current (trends in) DAC designs possible by re-examining even really quite old digital audio technology, ie going back to 70s techniques.
Others may disagree, of course, but AN DACs with no oversampling often seem to be well regarded among certain groups of listeners. So clearly thereās some mileage left in āoldā technology like 16/44 and so on.
1 Like
No doubtā¦ the algorithms have been reasonably well established over decades, the game changer in my opinion is the advance in extreme low power processing.
Traditionally increased processing in the DAC chain has generated heat and EM noise (both things that need to avoided/carefully controlled), as this can introduce perturbations/distortions which can undermine or cancel out the advantage of greater processing (particularly in filter reconstruction).
Now we have extreme low power processors and FPGAs, these processing bettermentās can now be applied without being cancelled by the distortions that otherwise would have occurred in earlier componentry.
Ironically to get the best out of 16/44.1 you really need state of the artā¦ less so for Hidefā¦
1 Like
For DSD, I am not quite sure of its benefits.
I believe the encoding for DSD is 0=go lower and 1= go higher. Theoretically, it takes a 16/44.1 can go from 0 to the max 65535 in any one sample, but it takes 65535 samples for DSD to do the same. So yo translate the sample frequency for DSD to 16bit you need to divide it by 65535?
So the sampling rate is really not that high?
Besides, supposed the value need to hold to the same, PCM encoding can make it the same but presumably DSD will need to have 101010 to +1-1 generating high frequency artifacts?
a common view in pro-audio. going 88/96 is beneficial mostly for semi-pro gear.
many stick to 44 to avoid sample-rate conversion and better compatibility with existing equipment in the studio. the money is still in 44 (CD, MP3, AAC). or 48 for video. you also need extra space to store masters at 88/96.
I do part-time DSP-development (software instruments) and dropping support for 44 is not even on the map. if I need more for some algorithm I just upsample
for low-bandwidth productions (jazz/classical) it probably looks different.
there is also confusion - what do a consumer that buy 24-bit want? the dynamic range of 16-bit is not used today. in most domestic playback there is limitations (social) how loud you can play and without reducing the dynamic range low-level detail (nuances) would be masked by the typical background noise in residential areas (25-30dB). an LP has typically 75dB range and sound fantastic with decent hardware.
if you release a recording with full dynamic range and the listerner cannot listen at natural levels (almost no-one can) and reduce volume will sound flat and boring due to loss of detail. like audiophile recordings often do 
AIUI 24bit reduces Quantization Distortion rather than giving a bigger dynamic range.
Even low sampling rates with 24bit sound better to me.
What is also really needed is some kind of declaration from online-services what processing they do to the music files they receive.
If you take a master that has noise-shaped dither applied it should not be touched for level normalization (not even a digital volume control in the preamp).
Personally I stick to ripping CD:s until online distribution grows up ā¦ hopefully Apple jumps in soon.
Go bust would be my preference. 
2 Likes
I think youāll find most internet streaming businesses simply distribute the master provided them by the distributor after all audio and video streaming is a pretty mature business operating modelā¦ the days of the original Napster approach are long oneā¦ ā¦ so really what would more appropriate is what processing is done on these distribution masters whether streaming, video audio streaming, CD, radio playout or vinylā¦ by the distributor ā¦
I have compared a Tidal master with a CD I had of the same distribution master, and I could see or hear no difference (once transcoded to LPCM)ā¦ admittedly I did this for only one masterā¦ but that was enough for me.
1 Like
As far as I know the different streaming services tells you what they want. If I remember this is was started by Apple with the āmastered for iTunesā program. Today Netflix, Spotify and others all have their own set of requirements (level in LUFS etc.).
So it is up to the mastering (in best case) to deliver. Otherwise the streaming service normally want a WAV and they will run it through their automatic process.
I dont know about Tidal and Qobuz.
But I see this as an area where CD and streaming are different. With CD the delivery is a 44/16 which no-one touches ā¦ it goes directly to DAC and then analog preamp
But in streaming people are doing all sorts of stuff to the digital data. Various corrections, digital volume, upsampling, transcoding
Trying to understand streaming, this is on the Streaming Audio section:
āā> Playing files stored on a Mac Mini directly connected to a pre or DAC I would not call streaming, as no network is involved.
"Being a novice, I would think that the above procedure would be the purest route!
WAV, FLAC, or non-compressed apple files ripped from my MM via itunes.
MM > Ethernet > Airport Extreme > Ethernet > Airport > Express > Ethernet > Inexpensive Arcam CD/SACD/DAC."
Where is my thinking gone flawed?
Please helpš
It just appears that a direct connection would be purer than the wireless connection Iām using now!
Plus I could utilize my CD, SACD, and stored local library on my MM for the family room!
And Living room system non-streaming:
CDX2>282>HCDR >Tri-Amped Briks
I did as Simon had done in the post above but with Qobuz as an additonal comparision. It was the same as streaming from a local source with 16/44. Iāve also compared a 24/96 album downloaded from HRA to the Qobuz hires stream and again it was the same despite being from different āsourcesā.
I donāt think the Music streaming services want WAV as they would have to convert to FLAC before they stream it to users. I donāt know what format gets sent to to Tidal, Qobuz etc
Transcoding from FLAC to WAV changes nothing other than unpacking the compressed data back to its original format.
Yes but other hi res like DSD increases quantization distortion. So it really is 6 of one, half a dozen of the other.
Iād agree with that. I find that on really low end hi-res gear like iPod docks that support it, they get an instant boost in performance from it. Either they are rigged to detect it and expand the dynamic range or it is just easier for cheap gear to digest. Then you up the quality of the gear and the differences start to melt away a bit. At least the really in your face differences that are heard on the bargain basement stuff. They are always doing hi-res a/b comparisons from the same master on shop floors here with the nastiest gear you can find and it really does sound different.
1 Like
What frustrate me is I want them to say so. At least for the 44/16 bits. āThis is the CD master bitsā. I fully understand this is meaningless for hires formats as there is nothing to compare to.
Then you only would have to make sure you local music server is setup so it does not mess with the bits with level matching, DSP-functions or upsampling.
If you listen to direct streaming from the net I am pretty sure all have a big automatic digital volume control in the signal path (I really hope they dither). They match levels. The services do not even measure it the same way ā¦ Tidal use LUFS while Spotify use ReplayGain and Apple has its own Sound Check Algorithm. EBU (radio/TV in europe) use LUFS at a completely different level to Tidal. And on it goes.
These days there are too many streaming services so you cant make one file for each service. You send one file to an aggregator who then distribute to all those services and the services use their software.
Given the state of the internet, if there ever was a case for blockchain. Here it is!
Mycelia Blockchain Project for Music and Rights
1 Like
Have you any examples of differing volume levels on Tidal vs Cd rip ?
Yes it may be different on lossy formats where a degree of optimisation is required to compensate for the loss in information in the encoding , but on lossless it appears it is simply the distribution masterā¦ and might even be the same masters used for CD production and possibly downloads. No eq or normalising etc unless in the distribution master, although I can add if required but I never doā¦ but if making a playlist from Tidal, I can find the mastering compression and loudness levels can really vary on different mastersā¦ so the lack of normalisation can be an issue, but I prefer it that way; ie Tidal not fiddling with the PCM. I also sometimes notice loudness level variation across different CD masters of the same trackā¦ so the master, version and distributor can be important.
However MQA certainly can mangle the sound, but I donāt tend to bother with itā¦ I canāt compare MQA with WAV LPCMā¦ as it is very differentā¦ certainly at the digital level.
Can anyone name a specific version of a track on Tidal or Qobuz where there are two versions of the same track at different sampling frequencies where they can hear a significant and clearly identifable difference between the two versions of the track?
If so, please let us know.
thanks
Jim
1 Like
Youāre probably best using something you are familiar with.
Sure, although same sampling frequency as Tidal lossless is all at 44.1
The Specials and one of my favourites Ghost Town
The masters on these two albums have different loudness levels and slightly different lengthsā¦ but it it is from the same mix master to my ears.
A Special Collection
and
More Specials
1 Like
Local streaming and streaming services unfortunately suffer from the interpolation of switch, modem .
Latest streamers may help the latter sound better compared to older gen. ones but just make a direct connection to a nas and you āll find where a bottleneck always lies.