Linux Audio the way to go!?

[soundcheck]

Hi klaus,
thanks for those very clear instructions.
you this is better than the RT kernel in the ubuntu repositories?
regards, stefan

ASOLUTELY. With the Gutsy rt-Kernel you are at 2.6.22.

Already in 2.6.23 a lot had changed and dramatically improved.
Dynamic Ticks, CFS scheduler and so forth,
With 2.6.24-ZEN you'll be already far ahead of the new Ubuntu-Hardy kernel, which will be released in April.

With the ZEN-kernel you'll also get the very latest ALSA revisions, which is IMO an important factor looking at the purpose I (we) use the kernel for.

Of course I have to say. If you want to try fancy 3-D stuff and gaming
on your PC you need to spend a bit more time to integrate the
graphic drivers again. Once you've upgraded to ZEN, you need to
reinstall your drivers. And you can't take the ones available from Ubuntu repositories.

Anyhow. Having a powerful 3D machine and compiz-fusion or similar apps. running and at the same time you like to have a powerful audio machine contradicts each other. This is what I like about UbuntuStudio. It's a "no frills" fast and nicely integrated audio solution.

If somebody want's to know how the reconfig of the graphics-driver would work, I could at least write it down for ATI graphics cards.

Cheers
\Klaus

 

[soundcheck]

soundcheck,
Thanks for the script, it worked well. I also modified your script to play one flac file at a time, instead of an entire directory.
My (modest) Celeron 900 needs just 5 seconds to copy a 15 minute song from flac file to tmpfs, before playback starts.
I experimented with tmpfs and ramdisk. There doesn't appear to be much difference, except that tmpfs will use swapfile if it's available (which is a bad thing in my opinion).

So brutefir is working great ...
but can it pause/resume?
Can it search forward/search back?
Maybe via keystrokes defined in brutdefconf?

I'm running Puppy Linux with 2.6.24 kernel patched with the Ingo Molnar RT patch.

Hi Linuxfan.

If you run brutefir as standalone application it doesn't allow lots of funny features.

You get runtime access to brutefir via CLI.
This way you can control at least the output volume and of course filter parameters .
If you play a list from a script you can kill the current process with Ctrl-C to switch to the next song.
That's about all luxury you can expect from a pure brutefir approach.

If you want to have fancy features try an application delivering these. You need to make sure that it streams raw data to stdout. Of course you'd need to set brutefir to stdin on the input. Afaik XMMS has a raw-stdout plugin. I havn't tested it by myself. You'd need to try it by yourself.
If somebody has accomplished this with Amarok let me know!

Cheers

 

[ecdesigns]

@ soundcheck

I hope you did not take my comments as an offence on your reclocker-module. I tried hard to point that out in my post
that I do recognize an improvement with your module..
Of course the 5V USB 2706 supply is an obvious weak point of
your concept.

no offence taken ,

I illustrated that PCM2706 jitter is quite high (higher than CS8416 SPDIF receiver for example), this is even the case when using self-powered battery power supply (2 x 1.5V AA alkaline batteries in series), and a 12MHz ultra low jitter clock.

However, since I used the 48MHz masterclock for both, driving the PCM2706 (48 / 4), and driving the reclocker, it was connected to the "noisy" USB power supply through a filter. This is not optimal, so you are still right about the power supply being a weak point, as it might affect master clock stability.

I already solved this "issue" by designing UTOS2 (USB + SPDIF / TOSLINK receiver with full galvanic insulation between source and DAC), and LRM2 (96 MHz local DPLL reclocker). Special power supplies are now used that enable performance similar to battery power supplies.

I am more focused on the different USB-Interfaces.

Results depend on the type of USB DAC used, if the USB DAC is sensitive to jitter, everything that even has the smallest effect on jitter (spectrum) will become audible. This includes moise, ground loop interference, 1 KHz USB bus clock stability, and packet size. I assume that highest USB bus clock stability, and equal packet size (if possible) will have a positive effect on connected USB DAC sound quality, when the USB DAC is sensitive to jitter.

In order to rule-out these kind of problems, High-End USB DACs must be immune to USB receiver jitter. Since masterclock jitter is also affected by (power supply / ground loop) noise, full galvanic insulation between both computer and DAC is also required. Transformers and opto-couplers still have a certain amount of coupling capacitance (few pF). Slaving a digital sound source often creates an additional unwanted galvanic connection (coax), and introduces more noise / interference.

TOSLINK provides both, full galvanic insulation (fibre optics), and zero coupling capacitance. TOSLINK induced jitter is of no importance here, since the DAC must be immune to jitter.

Under normal conditions, isochronous USB transmission errors are rare. If isochronous USB transmission errors do occur, they usually cause clearly audible drop-outs, or result in USB receiver lock-up.

BTW. Did you try the 48khz upsampling?

Yes I converted the 44.1/16 files to 48 KHz:

$ sndfile-resample -to 48000 -c 0 source.wav target.wav.

First conversion pass resulted in clipping, this was corrected during the second pass. Best Sinc Converter was used.

I used the DI8M that interpolates, and converts 44.1/16 to 352.8/19. The 44.1 KHz recording was played back at 44.1 KHz setting, 48 KHz at 48 KHz setting.

Compared to 44.1 native sample rate, the soundstage was no longer holographic, but rather "flat". I noticed similar problems with a North Star 192 transport upsampling from 44.1/16 to 192/24. Recent tests, using a new type of passive I/V converter, revealed this "effect" even more clearly.

 

[soundcheck]

@ ecdesigns

Interesting statements:

1. My Opticis fiber cable doesn't seem to be the worst solution after all to get a 100% galvanically isolated solution. ( I hope I get my TeddyRegs fixed over Easter to improve its PS)
2. I do believe that if you got your DAC pretty well in shape
the 48khz might be worse then native 44.1, if in both cases
the receiver is slaved to the DAC-Master.
However I can state ( and Eddie can confirm this) that in my
rig the 48khz did improve the seperation, pinpointing and soundstage quite impressive towards a holographic picture in
front of me (us). I wouldn't call that flat.

Let see what improvments I can achieve over Easter.

Cheers
\Klaus

 

[phofman]

If high-quality resampling from 44.1kHz to 48kHz improves the sound, I would check if there is no automated conversion 44.1kHz to 48kHz along the way to the DAC (e.g. dmix, hardware conversion in the card chip). This kind of online conversion is of lower quality than proper sox polyphase etc.

 

[ssmith]

If high-quality resampling from 44.1kHz to 48kHz improves the sound, I would check if there is no automated conversion 44.1kHz to 48kHz along the way to the DAC (e.g. dmix, hardware conversion in the card chip). This kind of online conversion is of lower quality than proper sox polyphase etc.

Hi, I have also tried offline upsampling with the Secret Rabbit Code to 48khz from 44.1khz.

However, I would not describe it as a dramatic difference. At first, with casual A/Bing, it seemed to be different, not necessarily better. Then I thought it was much better.... and extended comparisons gave me the impression that the sound had a little more top end, and the the soundstage was just ever so slightly improved, although not massively so.

Like Soundcheck, I am also using a DDDac. So this effect could be DDDac specific. All the settings were for no software upsampling to be done, but with computers one never can tell quite what they are doing.....! And I would not be at all surprised if offline upsampling had a negative effect with some dacs, and positive with others. So my flacs are backed up as original 16/44.1, and those on the hard drive for listening are upsampled to 16/48.

cheers, ssmith

 

[soundcheck]

If high-quality resampling from 44.1kHz to 48kHz improves the sound, I would check if there is no automated conversion 44.1kHz to 48kHz along the way to the DAC (e.g. dmix, hardware conversion in the card chip). This kind of online conversion is of lower quality than proper sox polyphase etc.

No. There is no upsampling in between. I think I have my Linux setup under control.

One thing for sure. Transfering 44.1 over USB is causing quite some non-linearities compared to 48. The 12 MHZ clock in relation to 44.1 are not n*x.
It is a kown fact that 44.1 can not be evenly distributed over the
USB frames. There might be other reasons.

If the ecdesign recklocker can handle this - great. That's why I am trying it. A standard PCM2707 receiver without reclocker cannot handle it properly - this is a fact .

As I mentioned before. I liked my earlier setup with 48khz better then a Benchmark DAC1 (44.1 or 48 material )and still better than my current setup on 44.1. The better your system gets the more obvious you'll hear the differences.

Let see what the weekend brings. I just picked up all parts for the regulators.


Cheers

 

[phofman]

To ssmith:

Thanks for you info. I did not mean conversion from 44.1 to 48kHz improves sound. On the other hand, every upsampling, especially of non-integer ratio, introduces distorsions. I mean that IF upsampling in SW improves the sound, it is very likely there is some low-quality internal conversion which automatically kicks in for 44.1kHz signals. E.g. dmix or many audio cards running on 48kHz internally.


To soundcheck:

Well, the USB argument could be the case. I think adaptive USB sound card is no way forward in audio engineering.

 

[phofman]

To soundcheck:

By using proper PCI (i.e. asynchronous) card (and perhaps using I2S, either DIY or the proprietary TG link), you could spare yourself most of the struggle with RT, low latency, loading to RAM etc. etc. Standard correctly setup linux would play just as good. For recording, it is a different story, as you need synchronized recording/playback, low delays (e.g. with MIDI recording), and the RT stuff makes sense.

 

[soundcheck]

To ssmith:

Thanks for you info. I did not mean conversion from 44.1 to 48kHz improves sound. On the other hand, every upsampling, especially of non-integer ratio, introduces distorsions. I mean that IF upsampling in SW improves the sound, it is very likely there is some low-quality internal conversion which automatically kicks in for 44.1kHz signals. E.g. dmix or many audio cards running on 48kHz internally.


To soundcheck:

Well, the USB argument could be the case. I think adaptive USB sound card is no way forward in audio engineering.

It's a bit complicated to convert the reclocker to 48khz. I am still
planning to give it a try, to see if I'll end up at the same result as reported by ec-design.

I am also well aware that the 44.1/48 conversion induces slight problems.
As usual. At every stage of your audio evolution you need to ask yourself: "What's the less of two evils?"
48kHz was the much better choice at my "DDDAC" evolutionary stage. ( BTW it's not only the DDDAC. A TwinDAC showed the same improvements.)

Let see what the time brings.
I am counting on full digital amps. No traditional DA conversion in the chain at all. Just PCM2PWM - that's about it. That's gonna be
my next project. There are just two threads about it over here.
One about tweaking a Panasonic XR59 or similar and Koonxxxx
who built the AFAIK first full digital DIY amp, that I am aware off.

For my current setup, a PCM2707 based DAC, the "RT-Struggle"
makes pretty much sense though. I'd guess that 80% of all DIY-DAC setups are exactly based on this jittery basis.
Who is able to or who is willing to spent thousends of $ for highly pimped DACs? - Just a very small minority - fair enough!
I am just showing an alternative that with an hour of configuration and a 150$ DAC the majority might end up at 1000$ Benchmark DAC 1 sound level.
Of course I could buy me and AES16 and some TACT amps. I am well aware of the theory.
But that would be too easy, wouldn't it. And I wouldn't be happy
about it.
Sometimes you need to try different things than the mainstream, especially if you're in the DIY arena - that's the real fun about it being around here.

Cheers

 

[phofman]

Well, it's certainly all about having fun

If adaptive USB is suboptimal for 44.1kHz and requires CPU intensive resampling, another standard way (in DIY too) is using a PCI card with SPDIF output and DDDAC with SPDIF input. SPDIF is technically inferior too, but PCI asynchronous communication would shield it from the possible software-induced jitter. Costs would be comparable.

Anyway, thanks for the great job.

 

[soundcheck]

Hi folks.

Have a look at Ubuntu Studio 8.04 Beta.

It will be the best off-the-shelve Linux setup for Audio purposes you've seen so far.

http://cdimage.ubuntu.com/ubuntustudio/releases/8.04/beta/

I strongly recommend to try this one. Have fun!

In April they'll launch the official release. I guess, once you got the beta installed, it just needs a software update to get the official release going. It won't be waste of time to install the Beta now.


BTW:

I finished my DAC tweaks for now. Even though I am using a ec-designs reclocker, Teddyregs, Opticis Optical USB cable and so forth, the quality of the incoming datastream received from the PC is still making a substantial difference.

As we all know Jitter is a cumulative beast.
Conclusion - supporting my very early assumptions - it's best to avoid jitter at any stage of the chain.

Cheers

 

[ssmith]

Hi folks.

Have a look at Ubuntu Studio 8.04 Beta.

It will be the best off-the-shelve Linux setup for Audio purposes you've seen so far.

http://cdimage.ubuntu.com/ubuntustudio/releases/8.04/beta/

I strongly recommend to try this one. Have fun!

x2
The amd64 version is now up and running on my audio PC -- very impressive, and definitely the most painless Linux install I've ever experienced, far easier than XP in fact. Even my wireless is working

 

[mafro]

Hey all, im new to the whole Linux audio thing and have spent a fair amount of time messing around with different headless audio players to run on my server. MPD and pitchfork seems like a pretty safe bet for now.

However, im more interested in achieving audiophile quality in the long term - which is why I just spent about 3 hours reading through this thread! Very interesting stuff all round guys, and very informative too..

USB Audio sounds like a safe way to go, and the observations from soundcheck about the digital amp thing are interesting, but for the time being I just need to get an outboard DAC to link upto my soundcard.

Ive got an old Audiophile 2496 from a few years ago when I was making music on my windows box, and it works fine under ALSA. The problem of syncing the S/PDIF can be handled with this card by getting it to sync to an incoming clock on the S/PDIF in. Does anyone know if the ALSA driver supports this?

If it does, can anyone recommend a couple of decent low to mid-range DACs which would produce a clock signal my card could sync to? Feel free to shoot me down if this is not the way to go.. Ive been sending S/PDIF to my old Sony minidisc deck and using that as an outboard DAC for many years, but im sure I can do better for quality

Also does anyone have any other experience with this soundcard? Im not against the idea of binning it and going another route.

Thanks for any help! Hopefully I can contribute something in the future..
mafro

 

[anbello]

@mafro:
If the Audiophile 2496 is based on Envy24* (i think it's true) you can set the board to sync to the incoming s/pdif signal with alsamixer. I have a Terratec Aureon 7.1 Space (based on Envy24HT) and i can do this.
Then you need a DAC with a masterclock and use this to send an s/pdif signal (with all zero data) to the Audiophile 2496. I'am doing this DIY starting from this post:
http://www.diyaudio.com/forums/showthread.php?postid=992869#post992869

Ciao
Andrea

 

[mafro]

Thanks anbello, ill keep my eye on that post as well. Ill also post any results I find out about this soundcard.

Whilst reading through that thread you posted, I came across a link to a Lithuanian company who make a nice looking DAC with a masterclock. They advocate a CD transport as the source, and I found this page on their site, which may be of interest to people here:

http://www.lessloss.com/computer_audio_usb.html

mafro

 

[ssmith]

If it does, can anyone recommend a couple of decent low to mid-range DACs which would produce a clock signal my card could sync to? Feel free to shoot me down if this is not the way to go.. Ive been sending S/PDIF to my old Sony minidisc deck and using that as an outboard DAC for many years, but im sure I can do better for quality

Hi Mafro,
What kind of budget are you aiming for?

 

[ssmith]

I am just showing an alternative that with an hour of configuration and a 150$ DAC the majority might end up at 1000$ Benchmark DAC 1 sound level.

Hi Klaus,

...an hour! I admit this Linux journey has taken me significantly longer than that...

But the way, was thinking that with the Ubuntu Studio 8.04 (Hardy Heron) release later this month, it would be a good time to do a major revamp and update of the wiki (or otherwise do a separate one) and base it around that distribution. The obejective would be to get something layed out for a total noob to be able to have a setup + key tweaks (Zen kernel etc) running as painlessly as possible. Seems to be a good time to do it, as 8.04 will be a LTS (long term support) release. On top of that, I'm really very impressed with it and reckon it must be one of the most user friendly distros out there.

I can try to get started on it later in the month?

regards, Stefan

 

[soundcheck]

The Wiki update is under production (offline). I got 60% finished.
I'd really like to open a Wiki somewhere else. I am not really happy with the structure over here.


NOTE: I am experiencing currently serious problems with the latest
automatic software update/upgrade ( introduced on the 2nd of April) of Ubuntu Studio 8.04. Beta.
X won't start anymore afterwards.
I installed the stuff the 2nd time now getting the same result.
I couldn't find any hints on the net yet.
So - folks be careful here not to waste time for now.

Cheers

 

[ssmith]

The Wiki update is under production (offline). I got 60% finished.
I'd really like to open a Wiki somewhere else. I am not really happy with the structure over here.

True, the wiki structure here is far from ideal, though would be good to keep it here if at all possible. What alternative are you thinking about? Over on the Ubuntu site perhaps?
If you need a hand let me know... I can at least help on the editing (that's my day job)...

NOTE: I am experiencing currently serious problems with the latest
automatic software update/upgrade ( introduced on the 2nd of April) of Ubuntu Studio 8.04. Beta.
X won't start anymore afterwards.
I installed the stuff the 2nd time now getting the same result.
I couldn't find any hints on the net yet.
So - folks be careful here not to waste time for now.

I got through that update unscathed....
I think the problem was due to the RT restricted driver package not being ready along with the new kernel, leaving your video card without a driver.
Maybe if you do an update from the command line it should repair itself. Otherwise check at the ubuntuforums -- I've seen a few other people reporting the same problem.
Still, best to steer clear of 8.04 until release. I'm not complaining though - I've had more luck with the 8.04 beta that with 7.10 and 7.04.