My 2 cents.
Computer or network based audio is extremely complex.
There are hundreds of sources causing this or that problem.
You solve one issue and you'll catch another one.
Every software and software update, no matter if we talk OS or application or device driver or device firmware or any change on the HW side can
cause new problems.
Comparisons between setups and devices are usually confusing or misleading. Most of the reviews about the subject you see on the net are questionable.
If you read the term "audiophile" in the context of OS or SW player. Stay away from it.
I started this thread in 2007. Talking about system optimizations and SW impact. We are still at the same point. The difference is that many commercial and also non-commerial apps integrate these optimizations
nowadays.
The key problem is that the DAC manufacturers do not manage to get the
mess under control.
If you're interested to read a recent and lengthy article about the subject, you'll find one over at audiostream.
I like the comment section most.
There you'll find a comment from
Gordon Rankin, the owner of Wavelength audio (high priced and quite well regarded dacs).
This guy is connected to many of the folks in the US high-end audio industry. And also sneaks around over at Audio Asylum.
After not being able to measure anything, with most expensive PRISM measurement gear, he came up with this conclusion:
He ran out of ideas, what to do next on the HW side.
They all talk hours and still don't nail the subject.
Hint:
To get around the issue of flac vs. wav on the SW side, you just look for a player that 1. decodes a flac prior to playback and stores the result in RAM and streams from there.
There are a tons of workarounds for other issues.
However.
Looking at the Metrum DAC. The digital interface they use, the M2Tech Hiface, is not the best in class SPDIF interface.
People over at Computer Audiophile rate the upper class Hiface EVO around 90% out of 15 interfaces. The Hiface will not be rated higher I'd guess.
Obviously this is again a highly questionable shootout. Anyhow. It gives some indications at least.
Basically the Metrum DAC is limited by the chosen internal OEM SPDIF interface.
Unless they'd introduce a highly sophisticated reclocking (see e.g. IANs project over here ) into their DAC, you will have an issue with this DAC.
Any changes/optimizations upstream will impact its performance. You'll again be gambling for best sound all the time.
What magnitude of impact are we talking about?
It depends on many factors. Person, expectations, bias, system, environment, ...
Can you hear any of this at all?
99% of the contributors of this thread since 2007 do not hear any differences, with any SW optimization. I and many other did and still do
hear a difference. You'll find the full spectrum of everything.
What to do?
I'm not aware of any DAC that has solved the main issue.
That's why you'd better stay with the CDP.
You'd solve most of the issues with one shot.
I for myself gonna stay with LogitechMediaServer, Squeezelite and my good old tweaked Squeezebox Touch feeding my 100$ full digital amp... ...for now.
My current main focus stays on the quality of the controller app, rather then slightest optimizations on the sound side. Orange Squeeze on my Droid tablet is all I currently see and use. (My system is hidden in a small cabinet and can be powered on/off via Wifi.)
The other day I made a friend aware of USB Audio Player PRO . A Android app. This app is using its own USB driver and passes the Android mixer by.
Ten minutes later he hooked up his Nexus 7 via OTG to his NAD M51 DAC .
Guess what. He told me it sounds pretty damn close to our tweaked Squeezebox setup.
Yep. Android a Unix system. They introduce mixers and samplerate converters to be able to manage the audio part of such a complex device. At that point they are not any different than Windows. Yep. There are ways to get around most of the trouble. Being bit transparent is the first challenge.
It again requires a reasonable app that does the optimizations.
BTW. Volumio (formerly known as Raspify) is meant for small embedded boards (RPI, UDOO, BeagleBone...). It's main audio engine is MPD, with all its issues and limitations. I recently did compare a squeezelite setup and mpd on a Raspberry PI:
All these little ARM boards can and do suffer of firmware/kernel and HW issues.
Those images available however are IMO a good starting point for rather decent quality playback.
Keep in mind though you're usually opening a can of worms, if you look for optimizations or specific setups.
Enjoy.
Cheers
Computer or network based audio is extremely complex.
There are hundreds of sources causing this or that problem.
You solve one issue and you'll catch another one.
Every software and software update, no matter if we talk OS or application or device driver or device firmware or any change on the HW side can
cause new problems.
Comparisons between setups and devices are usually confusing or misleading. Most of the reviews about the subject you see on the net are questionable.
If you read the term "audiophile" in the context of OS or SW player. Stay away from it.
I started this thread in 2007. Talking about system optimizations and SW impact. We are still at the same point. The difference is that many commercial and also non-commerial apps integrate these optimizations
nowadays.
The key problem is that the DAC manufacturers do not manage to get the
mess under control.
If you're interested to read a recent and lengthy article about the subject, you'll find one over at audiostream.
I like the comment section most.
There you'll find a comment from Gordon Rankin, the owner of Wavelength audio (high priced and quite well regarded dacs).
This guy is connected to many of the folks in the US high-end audio industry. And also sneaks around over at Audio Asylum.
After not being able to measure anything, with most expensive PRISM measurement gear, he came up with this conclusion:
Note: John Atkinson=Stereophile magazine; Charles Hansen=Ayre Audio
He ran out of ideas, what to do next on the HW side.
They all talk hours and still don't nail the subject.
Hint:
To get around the issue of flac vs. wav on the SW side, you just look for a player that 1. decodes a flac prior to playback and stores the result in RAM and streams from there.
There are a tons of workarounds for other issues.
However.
Looking at the Metrum DAC. The digital interface they use, the M2Tech Hiface, is not the best in class SPDIF interface.
People over at Computer Audiophile rate the upper class Hiface EVO around 90% out of 15 interfaces. The Hiface will not be rated higher I'd guess.
Obviously this is again a highly questionable shootout. Anyhow. It gives some indications at least.
Basically the Metrum DAC is limited by the chosen internal OEM SPDIF interface.
Unless they'd introduce a highly sophisticated reclocking (see e.g. IANs project over here ) into their DAC, you will have an issue with this DAC.
Any changes/optimizations upstream will impact its performance. You'll again be gambling for best sound all the time.
What magnitude of impact are we talking about?
It depends on many factors. Person, expectations, bias, system, environment, ...
Can you hear any of this at all?
99% of the contributors of this thread since 2007 do not hear any differences, with any SW optimization. I and many other did and still do
hear a difference. You'll find the full spectrum of everything.
What to do?
I'm not aware of any DAC that has solved the main issue.
That's why you'd better stay with the CDP.
You'd solve most of the issues with one shot.
I for myself gonna stay with LogitechMediaServer, Squeezelite and my good old tweaked Squeezebox Touch feeding my 100$ full digital amp... ...for now.
My current main focus stays on the quality of the controller app, rather then slightest optimizations on the sound side. Orange Squeeze on my Droid tablet is all I currently see and use. (My system is hidden in a small cabinet and can be powered on/off via Wifi.)
The other day I made a friend aware of USB Audio Player PRO . A Android app. This app is using its own USB driver and passes the Android mixer by.
Ten minutes later he hooked up his Nexus 7 via OTG to his NAD M51 DAC .
Guess what. He told me it sounds pretty damn close to our tweaked Squeezebox setup.
Yep. Android a Unix system. They introduce mixers and samplerate converters to be able to manage the audio part of such a complex device. At that point they are not any different than Windows. Yep. There are ways to get around most of the trouble. Being bit transparent is the first challenge.
It again requires a reasonable app that does the optimizations.
BTW. Volumio (formerly known as Raspify) is meant for small embedded boards (RPI, UDOO, BeagleBone...). It's main audio engine is MPD, with all its issues and limitations.
I recently did compare a squeezelite setup and mpd on a Raspberry PI:All these little ARM boards can and do suffer of firmware/kernel and HW issues.
Those images available however are IMO a good starting point for rather decent quality playback.
Keep in mind though you're usually opening a can of worms, if you look for optimizations or specific setups.
Enjoy.
Cheers
Thanks for the reply soundcheck.
One clarification, I've used my Audiophilleo 1 with PurePower over S/PDIF into the Metrum. This was compared to the CD transport also over S/PDIF. This takes the Metrum USB interface out of the equation.
Later I tried the Metrum USB input which, as you say, is a modified HiFace. The Metrum USB interface sounds similar to the Audiophilleo over S/PDIF.
The Audiophilleo seems to get favourable comments.
One clarification, I've used my Audiophilleo 1 with PurePower over S/PDIF into the Metrum. This was compared to the CD transport also over S/PDIF. This takes the Metrum USB interface out of the equation.
Later I tried the Metrum USB input which, as you say, is a modified HiFace. The Metrum USB interface sounds similar to the Audiophilleo over S/PDIF.
The Audiophilleo seems to get favourable comments.
LINUX vs Windows
Conclusion of my tests :
For linux systems in the best performance was with Oss4 drivers .
Achieves greater detail or depth in sonic detail .
The Alsa drives also provide good detail . The Pulse Audio in Ubuntu 12:04 greatly improved , however audible strip detailing. I usually remove Pulse Audio and use direct access ( HW : 0.0 ) .
Speaking of Windows:
The virtue of Windows 98 , which he wore was " governed " by the clock routines generated video card ( circuit based on crystal ) which gave stability clock ( 44,100 and 48,000 very stable ) .
I remember the first Windows XP there was a bug that the drivers got lost in your virtual reference clock . This is was a " pitch " in the tuning of instruments and even in BPM . Then came the " KB " to fix the " bug " .
Really on Win XP the sound of bass drums and Bass reveals more centered and " hard" . A good experience is to use the Monogram Graph Studio software allows you to route the Drivers and Dlls through a GUI . There are also other software that allows you to change the priorities of drives , by choosing who should be charged . So you can use either Windows or alternative codecs FFmpeg codecs .
In Win 7 the sound is more " aerated ." I conclude that this is due to a bit of (no intentional) reverb just occurring in the internal mixer Kernel routines and also the drivers of Realtek HD Intel boards ....
Conclusion of my tests :
For linux systems in the best performance was with Oss4 drivers .
Achieves greater detail or depth in sonic detail .
The Alsa drives also provide good detail . The Pulse Audio in Ubuntu 12:04 greatly improved , however audible strip detailing. I usually remove Pulse Audio and use direct access ( HW : 0.0 ) .
Speaking of Windows:
The virtue of Windows 98 , which he wore was " governed " by the clock routines generated video card ( circuit based on crystal ) which gave stability clock ( 44,100 and 48,000 very stable ) .
I remember the first Windows XP there was a bug that the drivers got lost in your virtual reference clock . This is was a " pitch " in the tuning of instruments and even in BPM . Then came the " KB " to fix the " bug " .
Really on Win XP the sound of bass drums and Bass reveals more centered and " hard" . A good experience is to use the Monogram Graph Studio software allows you to route the Drivers and Dlls through a GUI . There are also other software that allows you to change the priorities of drives , by choosing who should be charged . So you can use either Windows or alternative codecs FFmpeg codecs .
In Win 7 the sound is more " aerated ." I conclude that this is due to a bit of (no intentional) reverb just occurring in the internal mixer Kernel routines and also the drivers of Realtek HD Intel boards ....
Hi.
* OSS4 got compatibility/driver issues. The last build is from 2008 afaik.
Let me know if I'm not up2date.
* Beside using hw:x,x as output device, I still recommend to go for an
preemtible kernel/rt-kernel to be able to elevate the audio process/thread.
People who run Ubuntu can start with a lowlatency kernel, remove
pulseaudio (which makes use of preemption by itself if possible btw! ),
though this will break the audio control apps.
Daphile runs a rt-kernel btw.
It might be worth a try, for those who can live with a blackbox system (no ssh or login option).
From what I read the guy behind it seems to have focus on latest kernels, driver support, and yes, also SQ.
I just don't really like this "Audiophile" marketing BS about it.
Cheers
* OSS4 got compatibility/driver issues. The last build is from 2008 afaik.
Let me know if I'm not up2date.
* Beside using hw:x,x as output device, I still recommend to go for an
preemtible kernel/rt-kernel to be able to elevate the audio process/thread.
People who run Ubuntu can start with a lowlatency kernel, remove
pulseaudio (which makes use of preemption by itself if possible btw! ),
though this will break the audio control apps.
Daphile runs a rt-kernel btw.
It might be worth a try, for those who can live with a blackbox system (no ssh or login option).
From what I read the guy behind it seems to have focus on latest kernels, driver support, and yes, also SQ.
I just don't really like this "Audiophile" marketing BS about it.
Cheers
Last edited:
My first experience with Linux was in Ubuntu and Ubuntu Studio later on. In older versions it was easier to install Oss4.
Currently (2014 onwards), use Debian distributions such as Linux AP (not install Oss4) and AV Linux, which I consider the best distribution I've used so far (lets you install Oss4 without removing Alsa, and is easy to return the Alsa).
I've used Intel Realtek HD Audio, Maya 44 and EMU 1820M. M-Audio Fast Track USB (DAC).
The difference is in how Intel HDA is programmed. In Oss4 I directly access the HDA mixer, then just lower the gain and "mutate" inputs or adjust them as outputs. The mixer is intuitive.
Currently (2014 onwards), use Debian distributions such as Linux AP (not install Oss4) and AV Linux, which I consider the best distribution I've used so far (lets you install Oss4 without removing Alsa, and is easy to return the Alsa).
I've used Intel Realtek HD Audio, Maya 44 and EMU 1820M. M-Audio Fast Track USB (DAC).
The difference is in how Intel HDA is programmed. In Oss4 I directly access the HDA mixer, then just lower the gain and "mutate" inputs or adjust them as outputs. The mixer is intuitive.
Alsa uses Intel HDA facilities too, just like OSS. How does it influence the sonic performance? How did you do the comparison?
It's a while since I made any use of OSS4, but can't you just use dpkg-reconfigure linux-sound-base to swicth and stop the alsa service in Debian based distros? Something like this: http://crunchbang.org/forums/viewtopic.php?id=17428
Are new sound cards still coming on the market? phofman is well placed to tell me I'm wrong if ALSA soundcard devleopment is still very active.
My impression is that the focus of computer audio for end users has shifted from soundcards to external DACs with or without additional converters. USB audio class 2 and asynch mode is still held up as a benchmark. IIRC OSS4 doesn't support this and has been left behind.
My impression is that the focus of computer audio for end users has shifted from soundcards to external DACs with or without additional converters. USB audio class 2 and asynch mode is still held up as a benchmark. IIRC OSS4 doesn't support this and has been left behind.
Here is alsa-devel activity in January 2014: The Alsa-devel January 2014 Archive by thread . IMO the project is pretty alive.
External DACs are USB soundcards, usb-audio part is rather active. Lots of features for Intel HDA flow in constantly, new APIs for HW configuration etc. FW cards are slowly entering alsa area too. Not many new PCIe soundcards but that makes sense, Intel HDA codecs have improved a lot and lots of users go the USB path.
Of course majority of commits deal with embedded systems (the ASoC section) which reflects popularity of linux in the embedded world (and the lack of standards therein too).
External DACs are USB soundcards, usb-audio part is rather active. Lots of features for Intel HDA flow in constantly, new APIs for HW configuration etc. FW cards are slowly entering alsa area too. Not many new PCIe soundcards but that makes sense, Intel HDA codecs have improved a lot and lots of users go the USB path.
Of course majority of commits deal with embedded systems (the ASoC section) which reflects popularity of linux in the embedded world (and the lack of standards therein too).
Thanks for that link, and thanks for reminding me that External DACs are USB soundcards. It's interesting see where the activity is greatest. Talking of Intel HDA codecs on m/boards, what do you think of Gigabyte's amp-up audio, just a gimmick?
https://www.youtube.com/watch?v=GHjoTm_bhmg
https://www.youtube.com/watch?v=GHjoTm_bhmg
Well, [Phoronix] Linux Support Finally For Creative Sound Core3D does not give much confidence of proper support in linux. Like most other sound products by Creative.
I am assisting now a poor owner of a creative webcam who hit the same problem as kernel/git/tiwai/sound.git - Sound sub-system tree , just with a slightly different usb device ID. That hardware is a crap.
This latest release package (the 2008) installs on Debian from "xxx.deb" without recompilations. I learned the distributions as "AV Linux" (Debian) I just use the file available on the Open Sound website that installs the system alone prevents alsa loading ... and if I want to reverse the situation by command (root) I can uninstall OSS4. The ALSA driver returns to the Audio page.
The side effect after I return the ALSA was beneficial. I was not able to adjust the JACK Audio recording in 24 bit (EMU1820) and now I'm getting. Certainly some similar to the package that makes the pulseaudio was not reinstalled or changed configuration.
The side effect after I return the ALSA was beneficial. I was not able to adjust the JACK Audio recording in 24 bit (EMU1820) and now I'm getting. Certainly some similar to the package that makes the pulseaudio was not reinstalled or changed configuration.
The driver development process is detached from the userspace libs/tools now. The version numbering has been changed to follow the kernel version:
kernel/git/tiwai/sound.git - Sound sub-system tree
Alsa drivers git repo is maintained by Takashi Iwai within the official kernel git server:
Kernel.org git repositories
specifically kernel/git/tiwai/sound.git - Sound sub-system tree
kernel/git/tiwai/sound.git - Sound sub-system tree
Alsa drivers git repo is maintained by Takashi Iwai within the official kernel git server:
Kernel.org git repositories
specifically kernel/git/tiwai/sound.git - Sound sub-system tree
@soundcheck I believe that after 1.0.25, the linux driver updates are now committed to the main linux kernel, and it is no longer developed as a separate driver.
One of the developers explained this to me on the #ALSA IRC channel, I might not be saying it exactly right.
edit: ^what he said
One of the developers explained this to me on the #ALSA IRC channel, I might not be saying it exactly right.
edit: ^what he said
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.