Still waiting for you to send my 4 x 8TB WD Reds. Have you lost my address?
PS. Can cover postage if that’s the issue.
1 Like
Sure.
But it occurs to me, those four drives can hold over 50,000 albums, which cost $1,000,000, and take 25 years to listen to if you treat it like a full-time job. So are the drives really the problem here…?
No because like other people who live in the real world, I’ve got other things on them, like photos, video and apps. Because storage is free, apparently.
You are not seriously misunderstanding my statement, are you? Or objecting to it?
Yes, I absolutely think “cost is zero” is the most meaningful way to provide guidance to people considering technology acquisition design decisions.
“For hard drive prices, the race to zero is over: nobody won. For the past 35+ years or so, hard drives prices have dropped, from around $500,000 per gigabyte in 1981 to less than $0.03 per gigabyte today.” Backblaze.
“Zero” does not, of course, mean literally zero. Like in so many cases in computer technology, it means that the hardware cost is swamped by management cost, and retail and support costs, and packaging. And it means that for each type of data, the cost goes to zero literally, but it remains possible for us to invent new types of data that raises the demands by orders of magnitude: spreadsheets and word documents — unmeasurable; audio — vanishingly small; images — still small; high res video — yeah, it becomes interesting.
And it means that the cost of storage for music is a fraction of a percent of the purchase cost of the album, and about a percent of the cost of storage of the physical media.
I was kidding about the $1,000,000 worth of albums, but not really: the cost of storage for 50,000 albums, those four 8 TB drives, is $1,000 or 0.1 % of the value of the albums.
“Cost is zero” is the most meaningful guidance for technology acquisition design decisions, better than “cost is 0.1 %”.
As I’m suggesting in the previous post, I think this is a misguided optimization. 40 % of the storage cost for an album is 1 cent. For a thousand albums, that’s $10. Even thinking about and discussing the issue is wasting $10 in brain power.
“Storage is free” means “don’t bother optimizing it” because thinking is not free.
How do you enable this, my rips come as wav and would prefer flac since it takes up less space.
I believe there is a wizard these days built in for selecting FLAC, it’s been a while since I set up EAC. But also see https://wiki.hydrogenaud.io/index.php?title=EAC_and_FLAC.
I do a lot of listening with a friend who worked in the high-end audio industry for years and is well known in audio circles (including reps and manufacturers) for his ability to evaluate products for sound quality. He can easily pick out differences in sound quality that I sometimes find difficult to discern, so this is one instance where I know that individuals have different abilities to hear differences in components, software playback, etc. Sometimes it takes training to know what to listen for, and sometimes we have biases (tubes versus transistors, vinyl versus digital) that influence our listening preferences. We have had friends (audiophile as well as casual listeners) do some of the same comparisons in our systems with variable results – some people can quickly pick out differences, while others cannot.
There is definitely variable sound quality among available playback software. In our published testing of playback software and WAV/FLAC files (2-Part article in Absolute Sound Issues 217 and 218, and later 3-part article in Issues 246, 248 and 249), we found that iTunes had the poorest sound quality and that jRiver was easily superior in all respects. At the time of those published tests, Roon and Tidal were not available. More recently, using the same testing criteria, and maximizing each program’s settings for best playback, jRiver was easily superior to Roon using the same ripped WAV files and downloaded WAV or FLAC files from HD Tracks when compared over both software playback programs. In my system it only took a few seconds of listening to hear the difference. My system consists of B&W 802 D2 diamond speakers, PS Audio BK amp and Perfect Wave DAC, and a custom server running Windows 10. I enjoy the benefits of Roon for sorting music and the tagged info on the artists as well as the integration with Tidal, but for serious listening I much prefer jRiver for my ripped files.
Also, I still subscribe to Tidal Premium (CD quality) simply because I like the ability to stream music that is not in my home library and when I am away from home on my portable sound systems. Tidal (using Roon) is superior to other streaming services, but their streamed FLAC files are not equal to the same CDs that I ripped via dbpoweramp in FLAC format when played on Roon.
And I never hear a difference between Flac rips and Tidal CD quality in Roon. They can all sound great, so it just goes to show it’s in the brain of the beholder. As such, there is no right or wrong, you have to listen for yourself. High quality intimate Live music is my reference touch stone for this. The live recordings we make (with consent) often sound far superior to the CD versions in my opinion as it has an impact an immediacy that is very often lost in the studio. They have to make it great on the day and the best musicians can do this night after night.
POI we record ar 48/24 direct to disc and have 32 channels available. We often have 8 mics on Drums and use 2 room mics. Alongside
The sound on the night is a Mackie PA system with 4 active speakers time delayed front to rear and 2 subwoofers. The one in front of me can make my trousers flap lol
Also various Monitors
We have professional sound engineers run this lot lol
Thanks for taking the time to answer my questions. I’m no Absolute Audio subscriber so I couldn’t check what the test environment looked like. When you say
I must ask if it was so called bit perfect playback what was used for comparing? As soon as filters get involved things aren’t really comparable anymore.
What I was wondering about – and still am – is that the decoding of different lossless audio formats (WAV, ALAC, FLAC) results in different PCM data. This should be measurable and if there’s a difference the decoder doing it wrong could probably get fixed. I’m sorry but I’m still not seeing how this could be the case for either Roon or jRiver - because it could be either or even both of the programs getting the decoding to PCM wrong; it may even be one wrong sounds better for a listener than the right – or another wrong for that matter.
Measurable in the case I’m trying to make is comparing the PCM data after decoding by the respective software. At this stage hearing is not required. If there’s a difference between the data then there’s at least a chance it’s audible. If not the problem is not the codec / the decoder but something else in the chain.
Backblaze’s statement aimed squarely at the enterprise. It has almost no resonance in a domestic environment, where the cost of storage is accounts for a significant proportion of a home setup. My own 4x4TB disks cost more than than the NAS itself.
I’m glad you got there in the end because you continuously knock out the trope that it is when it patently isn’t true. Nor is it an insignificant consideration for anybody considering a new build system based around hard-drive sotrage and backup.
Huge fan of DBPoweramp, like the bit accurate feature so you know once ripped the disk is properly encoded.
Amazon sells 8 TB drives for $200, which is 2.5 cents per GB, slightly cheaper than Backblaze’s numbers. It’s actually the opposite: for a large enterprise like Amazon, Microsoft, and Google who spend billions per year on storage, the cost matters, but for consumers.
If you need 16 TB, let alone 32, you’re an outlier. Even 8 is high. A 4 TB drive is $100, with a life span of maybe 4 years, which looks pretty good compared to your internet connection, mobile service, Spotify, Tidal, Roon, Amazon Prime premium per year.
I have 1650 albums stored in 1 TB, which is $25, less than any DAC or brand name cable. So storage per album is less than 2 cents, which is 0.1% of the album, 5% of postage, roughly 1% of the cost of storing the jewel case (IKEA book case plus floor space in your house).
So I argue that disk storage is cheap enough that it is not something we should optimize. Or worry about.
I read an article a while back about the demise of CDs and DVDs, mentioning that burglars don’t bother picking them up. I wonder, if I had a few 8 TB drives lying on my desk, would a burglar bother to pick them up?
What you really mean is that storage cost is approaching zero – both in absolute and percentage of whole terms. So, why not say that? Approaching is the key word.
AJ
Yes, as an analyst of computer technology cost I would say that.
As a consumer of storage technology, I make the stronger statement, because it has already reached the point where it is not a factor. In the sense that if I choose to rip 1000 albums in both FLAC and ALAC, the extra cost is $20, which is not a deterrent to doing that.
In comparing Roon to jRiver, both programs are using bit perfect playback with the same WAV file feeding USB output to my DAC. JRiver has output options for memory and prebuffering that are audible; the settings do not affect the bit rate or bit perfect playback, but may reduce jitter, which seems to the factor that is influencing the sound quality. Regardless of how I configure the playback options for either ROON or jRiver, I still find jRiver to be the better program in every comparison.
In another finding, it is important to reduce or shutdown any programs that may be running in the background of the computer/server. Changing the priority of the playback program can make a difference (open Task Manager, click on the “Details” tab, find the program, right click, select “Set Priority” and change priority to High or Real Time. Everything done to reduce the load on the GPU seems to impact sound quality. In theory “bits are bits,” yet all these tweaks are probably impacting jitter and are audible in my system. I find that systems with lower resolution can obscure the audibility of such tweaks.
So this means a Nucleus built for the job with nothing extra running should have the best sound quality and also the quality of the end point has a major impact.
So a comparison of Roon and J River etc has to be done on exact same systems to be valid in any way.
Am I wrong?
Did you do any testing using a server/endpoint configuration or were all of the tests done PC to DAC directly?
I ask because I have tried jRiver, Audirvana, and Roon with a Sonore ultraRendu, all three setups feeding my PS Audio DirectStream DAC with no filtering or upsampling enabled. I would say that Roon and Audirvana sound the best with jRiver third. In a Mac direct to DAC setup, I like Audirvana best and Roon and jRiver about the same. I have not tried this with a Windows system. But, the server/endpoint setup sounds noticeably better than Mac direct to the DAC. This makes total sense as the ultraRendu is a custom designed computer made just to stream music out of a very clean and well designed USB port. There is nothing extraneous to lower the sound quality.
AIFF was not created by Apple; the A in AIFF does not stand for Apple but for Audio. It was created by Digidesign (now Avid) for ProTools and the like.
1 Like
Thanks again for the details. I understand that it’s not about the actual audio format / decoding but the implementation of data transport using Windows on the decoding machine. I can’t comment on this (said goodbye to Windows on private computing devices years ago) but after all
(1) it is not the decoder implementation causing differences but maybe other software tricks compensating for probable hardware deficiencies and
(2) as @Chrislayeruk mentioned since running the decoding Roon Core on a Nucleus / ROCK NUC is the most perfect way possible at the moment: one could compare a Nucleus with an jRiver IdNUC and listen how that sounds. That would be another story …
1 Like