 
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
|
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
37.201.195.134
As some of you might remember I own a RPi5 with 8GB RAM for about a month.
A very busy month. ;)
I just concluded a challenging project to get this thing properly going.
"Going" in a way that I finally got my custom version of RPiOS
Bookworm properly integrated and all my custom apps adapted to the new HW.
All with the obvious goal to outperform - soundwise - my well established RPi4/CM4 setups.
SW-setup:
I am running
* my own custom operating system - no pre-made image,
* my own custom realtime kernel,
* a custom Logitech Media Server and
* a custom squeezelite
* numerous optimizations
* as controller the excellent LMS-MaterialSkin plugin by Craig Drummond
Basically there's not much that's original Raspberry PI OS.
Some key items I'd like to mention about the HW:
* The new RPi5 runs at about double performance
* The board layout and parts have been much improved
* Power consumption at idle is about the same as the RPi4
* I get along with a GeeekPi passive cooling case @54°C and continuous 2300MHz.
Downside:
* The RPi5 allows for much less HW teaks at this stage.
* Due to cooling needs working with HATs is basically not recommended.
Additional HW:
* I use my Gustard A18 hooked up to USB3
* a Samsung SSD also hooked up to USB3 via SATA-USB adapter.
* Allo Shanti 5V/3A supply
* iFi USB filters on DAC and SSD
* IanCananda ShieldPi Pro (GPIO filter and buffer) attached via 90° GPIO adapter
Outlook:
* I just received the NVME HAT, which I'll test soon.
Conclusion !?!?
All I can say. To wrap it all up. That'll be the last post that I am mentioning the RPi4.
Yep. I consider the new setup quite a step up the audiophile ladder. There's no looking back.
From an audio perspective the new setup exceeds my expectations. And I consider it a major success.
Enjoy.
-----------------------------------------------------------------
blog latest >> The Audio Streaming Series - tuning kit pCP
Follow Ups:
Some photos of "the big clean up" after 24/7 and not quite 365 and 4.68TB of data incoming over mobile.
Yes it's a CM4 on the swiss carrier board.
Bit dirty that!!
More
This is a 3d printed case which has had a number of iterations and mods to get here.
Notice the Ian C dual sabre DAC above the CPU.
That makes it non-trivial to keep the whole pile cool, esp in summer and used for HD video.
Another shot inside to demo the problems with the "PILE" and how can it be kept cool.
I have made an airstream diverter in alloy plate so it's dropped 5C to 45C running.
Next will be a 3d printed "ventouse" to get the fan stream direct on to the CPU.
The 12V fan runs on 5V to give a gentle shhhh not a mad howl.
There is an issue on the board because of the huge pile of GPIO extenders to get the height up above the CPU heatsink...they sit against the back wall of the case.
They partially bloc the airflow out the back of the box, so again another mod follows to get the hot air out.
Finally after fiddling around for 2 days, creating a carrier slice and mounting+connecting the Sowter transfos.
Drilling plenty of 2mm holes.
Works real nice.
DSP from Henryk works glitch free.
Balanced line out, low noise, clarity excellent into 2, 60W amps*.
*1949 fully rebuilt and improved KT8 based monsters from USA......
Try doing that on a Rpi5.
We will see when next 40C summer weather arrives!
This was the solution adopted after I figured the bisikek transformers were wildly overpriced.
Sowter btw has been going for decades making transformers that just work in mostly mumetal.
All the more exotic hype for nanocrystalline stuff = more hype for imaginary effects, while in the meantime people like us are CREATING the recordings you are listening to....microphones, DPA, Neumann, Gefell etc etc through a studio mixer and ADCs full of opamps.
The only worry I have in my CM4 - on a high quality swiss made board btw is RFI.
RFI from the wifi, and RFI from the 4G - LTE M.2 modem.
Both sources are switched to external antennae, something they missed on the RPI5, and another stupid thing...
The CM4 runs off car /24V or battery power because it's able to be fed with 9-28V DC.
The waveshare CM4 (power thru 5V USB-C) and Rpi5 systems can't do that - so maybe you're just listening to PSU noise?
(I happen to know that cos I have 2 waveshare boards plus the official CM4 9-28V one and a retroflag 2 GPI with Kodi + headphone output which has an onboard 4hr lithium battery and a CM4 inside.
"The only worry I have in my CM4 - on a high quality swiss made board btw is RFI."BTW I thought the Swiss were known for making great watches. Oh wait, several of my Swiss watches have failed yet the Japanese watches keep on keeping on. Maybe your Swiss made CM4 PCB has too many Swiss cheese like holes in it?
Edits: 12/28/23 12/29/23
"I thought the Swiss were known for making great watches."
Swiss watches?
they do indeed make super watches...only about 90 mins from where we are now..
which is why of course I am wearing one of them now..(a rare blackface Longines auto) more than 45yrs old - which is brutally mistreated most of the time).
Maybe you didn't get your watches maintained by the right people?!
(I get mine cleaned and worked on in Russia.)
As for the Swiss TOFU board, not only is it by far the best IOT board out there, but being 2-3x of the price one expects and gets proper support + a superbly made product. They are releasing an even better one now.
I have just finished cleaning the thing of dust build up +it has done 4.7TB over 4g this year alone as a router, file server, DSP, + loads more tasks/database/docker and more.
The only thing that has gone on it in 12 months has been a CM4 chip which died.
In an attempt to diss anything of quality, all you manage to do is to look incredibly stupid.
"different types of SSDs were causing different sound signatures"
"This CM4 setup was/is "in audiophile terms" considerably worse compared to my new RPi5 setup. The RPi5 is simply a lot
better in many areas. "
Sorry this is just manifest rubbish.
How do the electrons know the difference between one path or another?
All these effects are purely imaginary.
I don't believe any of this "audiophile" stuff.
Fact is any computer is a noisy environment but I have a comparatively ancient (AKM) DAC in a studio PCMCIA card in a notebook, and it's really fine.
The dual sabre based system is also fine - much improved it can be said by using a pair of Sowter 3284 on the output.
This provides good isolation and no extra extraneous feedback loops.
Those transformers were used in studio desks.
"different types of SSDs were causing different sound signatures." Sorry this is just manifest rubbish.Not for the first time, soundchekk is right. Not only can different SSDs sound different in otherwise idential audio replay circuits but, contrary to perceived wisdom, my experience is that a 'spinning rust' HDD typically sounds better for audio than the several SSD I've tried. SSDs are very noisy beasts.
I stress "replay" because the output from such devices is, by definition, real time. If you don't know why that matters in audio, I'd respectfully ask you to do some homework before belittling others.
Different makes of HDD also sound different: for example, a Seagate 'Momentus' 2.5" HDD (NLA new but easily found 'reconditioned') sounds better than other makes I've tried, perhaps in part because its power draw is about 0.5a at 5v compared to most makes which typically draw about an amp. (They're pretty slow data-transfer-wise but easily fast enough for audio replay.)
How do the electrons know the difference between one path or another?
Very droll but, to pinch a phrase, "manifest rubbish". There's a wealth of literature on the effect of electronic noise on audio performance though, sadly, rather less seeking to correlate the nature of such noise with how it affects perception. It's a difficult topic to research and probably beyond the resources of manufacturers, let alone hobbyists.
All these effects are purely imaginary.
I researched SSDs vs HDDs in the audio replay context a year or two back but didn't report what I found on AA precisely because I couldn't face the tedium of dealing with the likes of the above.
But, for those who care to experiment, I'd recommend trying Elfidelity SATA filters. Also, try powering an SSD or HDD not from the system's 5v line but from its 12v line with a decent in-line 5v Vreg such as a 'New Class D' device. You might be pleasantly surprised.
Dave
Edits: 12/27/23 12/27/23
yea right "a wealth of literature on the effect of electronic noise on audio performance"...
Sure, on ANALOG audio maybe, but even then you would be pretty incompetent to be there after people like Rupert Neve eliminated most of that with high quality transformers.
However digital audio is simply a stream of 0s and 1s.
It's simply binary mate!
You can't change the code with noise added or subtracted.
Playing back digital audio is simply a bit stream with a buffer.
With camillaDSP you can ask it to show the bitstream and the buffering in pretty much real time.
So put it this way, if a DSP can cope with the ups and downs of CPU and memory use without a glitch, then ideas about SSD or HD or all the other buffers along the route is just more BS.
"Audiophiles" love talking BS, and they love to invent stuff that doesn't exist, to prove their "golden ears" are a zillion times better than mine.
Well come on then come to our place and do a double or triple blind test to tell us which mics we use, or see if you can hear the difference between high bit rate mp3, 16 bit audio and 24-96.
You will FAIL.
I can bet a case of new year champagne on it.
or see if you can hear the difference between high bit rate mp3, 16 bit audio and 24-96.
Too funny! Here's the 44/16 version of the opening track to the soundtrack of Rogue One called "He's Here For Us":
By contrast, here's the original studio 96/24 version:
Perhaps you cannot hear the difference!
...the same section of music. The "audio position" of the 44.1K/16 track shows 1 min 24 secs while the second image shows zero. Plus, the time runs of the two tracks are different. Then, the 44.1k/15 track shows 96K/24 on the side.
CD quality has around 96 dB of dynamic range while hirez can go to about 120 dB from a practical standpoint. Contrast that to analog tape or LP formats which typically struggle to get past 70 dB of dynamic range. There is no technical reason that a CD quality can't provide the dynamic range needed for uncompressed music playback.
Now, what the record companies and producers CHOOSE to do with respect to dynamic compression is a whole 'nother issue. If your two images do present a dynamic range difference between the two formats for the same recording, it is because they =wanted= to do that, not because it was due to technical limitations.
First of all, there aren't hundreds of remastered releases for that recent soundtrack. The time segment differences simply relate to adjusting the default view.
Hearing the difference between the master and the 44/16 version are profound. See my other response in thread for more on that.
I've got some really good CD content from labels like Telarc and Windham Hill. Just never as good as the best high sample rate 24 bit content. The world's engineers today really aren't fools as some suggest.
> > "The world's engineers today really aren't fools as some suggest."
No one said they or the producers were fools. They put out content that they think will sell -- mass market and audiophiles are two radically different audiences.
We have plenty of CD re-releases of the same album that vary widely in quality -- sometimes the newest "remaster" is the worst of the lot -- and all that happened without changing the recording format.
There are playback situations where too much dynamic range is an inferior listening experience -- for example in a car where the pianissimo sections vanish into the road, wind and engine noise.
That said, I still don't believe the two screencaps you posted show the exact same section of music even if they are from the same performance.
No one said they or the producers were fools
I reference the Estonian guy who challenged an inmate to...
see if you can hear the difference between high bit rate mp3, 16 bit audio and 24-96.
That's hilarious or truly sad for anyone to assert that lossy 44/16 is as good as it gets. Ask any recording engineer if they share that absurd belief see what they say. I met Jack Renner when I participated in the Telarc recording of ASO's Firebird . He told Dr. Stockham of Soundstream forget it if the good doctor was not able to increase the sample rate of his recorder above the CD standard. That's why the very first digital recordings were 50/16 and later ones in DSD. The 44/16 standard was fenced in from the outset with the limitations of the late 70s era 700 MB optical media available at the time.
I still don't believe the two screencaps you posted show the exact same section of music even if they are from the same performance.
You are welcome to create your own reality! While the original version runs a bit longer in the display, both start from the beginning of the track. Which in its uncompressed form can raise the hairs on your arm with the initial SLAMM! totally lost with the 44/16 version. If you have Tidal or Qobuz you can hear the latter version. You'll have to download the studio version from HDTracks.
View YouTube Video
I'll take your word that the screencaps are from the same section of the same track.
But you still seem to be missing the point that the CD version's more compressed rendering was an intentional editing choice made by the people who produced the CD. It was not =necessary= to compress it for the 44.1/16 format.
I'm not at home right now, but later this evening I have an experiment I plan to try with your track.
But you still seem to be missing the point that the CD version's more compressed rendering was an intentional editing choice made by the people who produced the CD.
Evidently, you haven't read this post in the thread.
I logged onto my Qobuz account and found both the CD quality and hirez versions of the Giacchino track you showed earlier. I captured the first 1 minute and 20 seconds of each version. The 44.1K version first, then followed by the 96K version on the right side of the image below. As you can see, they appear identical.
I ran some statistics on each of them in Adobe Audition. The quietest section of actual music occurs about 17 seconds into each track -- both are identical at about -34 dB from the 0 dB max. With CD's 96 dB dynamic range, this leaves 62 dB of range =below= the quietest section of music in the first minute and 20 seconds of this recording.
The only difference in dynamic range I was able to find occurred at the very obvious peaks -- the CD version peaks at -0.49 dB while the hirez version peaks at -0.38 dB. I'm not sure an extra tenth of a dB is going to add much to the startle effect. But, that is also not to say the hirez version won't have a bit more resolution, probably more due to the 24 bits versus 16 bits than from the higher sampling rate.
But this gets back to my original point on the images you posted -- what they did to your CD version was simply an artistic choice on the producer's or record company's part. In no way was it required by the CD quality file format. I've never said hirez can't be better than CD, but the example you posted has very little to do with the differences in file format and everything to do with the intentional choices made by producers and record companies.
to Zack.Note the new 44/16 version is the expanded version where a better choice was made. Years after the first take.
Edits: 12/28/23
-nt
learning that five years after getting that music and making those comparisons, Qobuz now has both and offers the extended version with new content!
That has got to rate as the single most stupid post I have ever seen on here.
(And you are using Audacity)
Congratulations!
Amateur.
it is Audacity displaying the difference in dynamics between the 96/24 master download and a 44/16 release of the same content.Amateur.
If that's how you characterize works by Disney made at the Sony Sound Scoring Stage , suit yourself!
Edits: 12/27/23
Who is to say the original wasn't compressed to get to 16/44? That compression alone will make it sound different.
I've done comparisons with unsuspecting people where they are playing the Spotify version of a song through Airplay into my kitchen system, and I'll have them stop theirs and play the same song at 24/96 from Qobuz onto the same system and they are always astounded at the difference.
Don't know about you, but I live in the real world of commercial recordings. I buy whats available- not based upon theoretical postulations.
While lossless 44/16 can come closer to what the studios actually use, content is usually compressed for the masses in that format. Along with compromises in transparency required by the brickwall filtering known decades ago. Call me crazy, but I prefer listening to the master resolution.
Lossy is a last resort if that's the only way the music is available.
"Note:
We're a small group of audiophiles over here running pretty similar setups"
yes you are a small group of nutters claiming to hear things that you can't.Proof of this..
"Call me crazy, but I prefer listening to the master resolution".
You have not the slightest idea what is "master resolution".
If you had actually worked in a professional recording environment you would understand a bit more.
It's not about 16 v 24 or 48 v 96, there's loads more to it than that esp as we are mostly in AES67 territory now/multichannel/latency/convolution.
Studios make stuff for people who live in their little bubble and like to write blogs about what imaginary stuff they think they can hear in there....then criticise male chauvinist groups like ASR for going down their own rabbit hole.
They wouldn't even know most of that stuff has been subject to compression and convolution on a massive DSP based scale, because they wouldn't recognise the artefacts.
Another rabbit hole is the "tube v solid state" arguments, and yet another about speakers or even LS3/5 v any version of it..
I get fed up with shooting rabbits....
"I live in the real world of commercial recordings. I buy whats available".
No you don't you live in your small imaginary bubble of what rubbish is put out there on a daily basis by the commercial entities that determine what you should hear (perhaps), and are dumb enough to pay them for it.
If you were involved with live recording and live stereo broadcast + all the immense problems involved in monitoring it (I won't even start on headphones!), then you might get it.
I won't even start on the problems with surround sound, - which btw I have done quite regularly and successfully...
As for the original posts, I am tearing down the Router/DAC/CM4 system for a bit of maintenance today, and to make up a proper solid installation for my Sowter transformers, which send balanced lines to the amp inputs.
What is more important to me, is to keep that ARM CPU cool when outputting HD video at the same time as running the DSP and the DAC.
That's a tough call.
It makes an amazing home cinema system under Linux 64, but in hot summer weather it's a struggle.
I don't doubt the CM5 will be a horror story to keep cool in those conditions, especially as it has NO wifi ext antenna and NO 9-24V PSU possibility.
A good idea to release in NH winter eh, so as nobody yet knows how hard it will be to keep cool!
btw I do have a version of the CM4 running ANDROID with sound out on HDMI.
If anything it's the best of all for video and glitch free mpeg audio.
I run an enormous heatsink on it, as it's on the official Raspi board.
No you don't you live in your small imaginary bubble of what rubbish is put out there on a daily basis by the commercial entities that determine what you should hear (perhaps), and are dumb enough to pay them for it.
????
I call total bollox on all this rubbish.
I know what I am dealing with here...
"I am a graduated engineer" and "audiophile"!
The we finally have this jewel "Different makes of HDD also sound different: for example, a Seagate 'Momentus' 2.5" HDD (NLA new but easily found 'reconditioned') sounds better than other makes I've tried"..
Then you start dissing stuff you know nothing about...like Digigram products which are used daily in radio broadcast throughout Europe, while my mine works in Linux btw with 4 channels of XLR output built in+24 bit audio....I actually use for DSP and calibration from 10hz - 20khz.. low distortion and ruler flat FR.
Used also for checking IMD and FFT.
All the effects you detailed are entirely imaginary.
When you have actually done some studio and live music festival broadcast in professional environment you can right blogs on the internet.
I don't have time, and we don't give a monkey about what hard disk or mem component is recording or playing back our mixes.
But you have also stated the one truth that otherwise gets lost in all of the rubbish about digital, it is the ANALOG noise generated by the drives/computers/interfaces that make the difference. But, if you isolate that noise, either through wifi or optical, it stays where it was generated. The noise does not travel through the air. Digital streams are pure.
When that wifi hits the final receiver, the receiver can itself be generating noise, but at some point the DAC needs to convert the digital and reject the noise through whatever filtering mechanisms it has. No doubt there can be issues with that too.
And at the speeds offered these days, audio is just a drop in the bucket and can be retransmitted multiple times to ensure that it is right. Only the final leg needs to be real time. 24x384 DxD is about 20mbps, my wifi is 1000mbps, so about 50x faster. Oh, it is always right because everything that hits my computer from the internet is always right. You never see garbled text on a screen, do you?
But, if you isolate that noise, either through wifi or optical . . . Digital streams are pure.
Aye, right. As pure as my heart. Other than in a (perfectly proper, indeed essential) conceptual sense, there is no such thing as a 'digital stream' absent the medium that carries it.
. . .at the speeds offered these days, audio is just a drop in the bucket and can be retransmitted multiple times to ensure that it is right.
I don't get your point. Can you clarify?
You never see garbled text on a screen, do you?
Garbled is not the same as distorted. The resolutions involved in monitor-displayed text are so low that you don't know if it is distorted or not.
At the much higher high resolutions used for printed text in the graphic arts sector (typically 1270 or 2540 dpi) yes, you can, you do and I have.
Dave
Does the noise travel through the air?
Does audio happen real time over wifi? No. The entire track is forwarded before it even starts to play. And if it doesn't pass the error check it is forwarded again until it is right.
Graphics arts issues are with original resolutions and recordings. Once the text is in the high resolution necessary does it change upon transfer? There is a difference between creation and replication. Is every digital recording ready for playback? The studio guys would argue that it isn't. But once they are done it doesn't change just because you play it one way or the other.
"Sorry this is just manifest rubbish."
I was wondering what education/background led you to this conclusion!?!?
"I don't believe any of this "audiophile" stuff."
Believe - that we do in church. I am a graduated engineer with more
than 4 decades of technical experience. Just to mention it.
The key success factor for evolutionary success is flexibility.
Open minded people = flexible people, do have a huge evolutionary
advantage over narrow-minded, prejudiced people. It's the
open-minded people that are in charge for progress. Otherwise we all would still live in caves or work with ancient AKM DACs or PCMCIA cards.
I btw wrote a blog article about the whole subject, perhaps you'll learn something.
https://soundcheck-audio.blogspot.com/2021/04/everything-that-matters.html
Enjoy.
-----------------------------------------------------------------
blog latest >> The Audio Streaming Series - tuning kit pCP
yea but you probably never ran a CM4 especially on M.2 memory (PCI-E).
I reckon a good CM4 install (they do run stable at 2Ghz) is considerably better than any RPi5, and using a good board, makes the pi5 look not such a great deal after all.
...they do run stable at 2Ghz
with the RPi3 / DigiOne used in the garage system. As a music player, there's simply no need for Cray supercomputer power. Media library lives on a Synology NAS for shared access for all the players around the manse.
To reduce power consumption, heat and noise, I declock mine to 800 mHz (leaving memory running at half that) and power using a 1A battery. ultraRendu used upstairs also requires only 1A.
Me too. My pc is running at 480 mHz and doing upsampling to 192k
Cut to razor sounding violins
Great that you bring it up.
I've been running a CM4 attached to a Waveshare IO-Base in my "main" system
for quite some time. And I've been using a 2230 NVME SSD. Since about 6 months
a 2TB! 2230 SSD. (BTW: different types of SSDs were causing different sound signatures). I tried all other storage options (SD,eMMC,USB-SATA) against it!
With above setup I've been using my IMO (still) excellent Allo Katana HAT DAC.
Why?? USB is rather crappy on these Waveshare IO boards.
The system ran @2200MHz. At times I was running even @2300MHz. All with passive
cooling of course.
And this setup had been my best RPI audio streaming and playback setup for quite some time!
And. I also own all other types of RPis. Just to mention it.
Bottom line.
Here you have it. Black on white. This CM4 setup was/is "in audiophile terms" considerably worse compared to my new RPi5 setup. The RPi5 is simply a lot
better in many areas. Just have a look at the specs for a start.
Keep in mind though. Not just the HW matters. Poorly integrated SW can cause
a lot of trouble. Since the RPi5 is a rather new device, there might be space for improvement on the SW side for most of the people out there.
As usual. How one judges or values the achieved changes is a pretty subjective
matter with a lot of variables at play...
Enjoy.
Note:
We're a small group of audiophiles over here running pretty similar setups. The opinions or - most of the time - experiences I am expressing over here are usually verified against several other systems and ears out there.
-----------------------------------------------------------------
blog latest >> The Audio Streaming Series - tuning kit pCP
| FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
||||||||||||||
This post is made possible by the generous support of people like you and our sponsors:
Top of Page ]
[ Contact Us ]
[ Support/Wish List ]
[ Copyright © 2024, Audio Asylum, all rights reserved ]
