HRT Music Streamer II with KDE: what's the best configuration?

[mmdanziger]

Hi, I'm new to this site and I apologize in advance if I've misposted this question. I couldn't tell whether it belonged in the software or DAC section.

 

Anyhow, I've recently acquired a Music Streamer II and am playing around with it on my Linux box (Linux Mint 9 KDE). I've never really tweaked my Linux audio settings so I must confess, I'm a bit confused about the whole pulse, alsa, oss, phonon business.

 

I've selected the Music Streamer as my main sound card in System Settings and am getting sound out of it. The sound is a bit flat, but it's very sharp and very clean. I'm hooking it up to a Naim Nait 5i and Revel F12 loudspeakers. I think that's probably just the sound of the device, I bought it as a partial replacement for my Meridian 506 which is in need of major repairs. I guess for $150 you don't quite the same sound as a CD player which cost 20 or 30 times as much new.

 

My question is what to do to get the very best sound out of my device. Lossless is a no brainer. But is there any difference, from a sonic perspective, between pulse and alsa...oss4? How does KDE/phonon affect that question? Using one player over another? Amarok clearly uses the phonon backend while mplayer is self-contained, right? The quality of the USB cable? My interconnects are fancy audiophile grade but the USB cable is just a generic Chinese factory piece.

 

Sorry if I'm dumping a ton of unrelated or silly questions, I'm new to the computer audiophile game and the pages I found discussing the differences between the systems were confusing and largely avoided the question of sound quality.

 

If anyone knows something about this sort of setup, I'd appreciate it.

 

[Richard Dale]

I would say the best Linux solution sound-wise, is to use MPD directly talking to ALSA. I tried that with my HP 2133 netbook running Kubuntu, but thought my MacBook with Mac OS X/iTunes sounded better. There are some threads in this forum about how I configured my HRT Streamer II+ to work with MPD. I wouldn't use Pulse Audio or OSS4, as they do mixing and resampling which you want to avoid.

 

Amarok and Phonon can use several different backends, and the Amarok guys think the VLC backend is the best. I tried the XINE backend with my AIFF files and it didn't work well - there were noises in between the tracks. Maybe FLAC would work better, but I haven't tried. Pulse Audio makes itself look like ALSA to a client application and so I would expect every Phonon backend to work with both Pulse and ALSA, when suitably configured. I'm not sure which of the Phonon backends would work with OSS4 though.

 

An audiophile cable can easily cost at least as much as the basic Music Streamer II, and so I'm not sure if it is a good idea to change your current cable until you've settled on the software to use and have got it working.

 

[mmdanziger]

First of all, thanks for your response and your helpful posts in other threads. Following your advice, I installed the latest mpd 0.16 alpha something from the trunk ppa and told it to output to my device with hw:1,0 (no plug needed) and it appears to be working fine.

 

I have one glaring problem and that is that 24 bit files are often afflicted with unending clicks. They sound almost like vinyl noise but more jarring more frequent and have more of a click sound than a pop sound. 16/44.1 files play fine and sometimes the 24/96 files play fine as well. I have not been able to establish criteria for when the clicks will appear and when not. Could these clicks be lost data from a bad usb cable? A re-sampling artifact? ??? I have not edited/created an asound.conf file yet. I have been wary to mess with alsa-wide settings as I have no idea what I'm doing there and don't want to impair the functioning of my on board ALC888 which I still want to use for web browsing, games and system notifications.

 

Is this a common problem? Does it point to a particular "weakest link" or should I be doing more thorough testing?

 

Thanks again, md.

 

[Miska]

On Linux, pulseaudio is equivalent to kernel mixer on Windows or CoreAudio engine on OS X. It does sample rate conversion, mixing, routing and volume control tasks.

 

It can also sit on top of ALSA and present ALSA and various other audio interfaces upwards.

 

There are bunch of configurables to adjust resampling quality and such in pulseaudio. But for quality you may want to bypass pulse altogether...

 

 

[1audio]

I found Pulse Audio to be a pain and judging from the Linux forums many others trying to get good audio find it to be a major bottleneck. If you use MPD its pretty easy to have it talk directly to ALSA. I would use plughw:n.n instead of hw:n.n, the overhead of alsa repacking the data is trivial and it will be correct for your card.

In a desktop system you may have issues with priorities and other process interfering with the audio. Linux gives you access to alter those relationships and millions of Google hits to wade through to figure out where they are and how to adjust them. If you are a dungeons and dragons fan its like a superset of the game, but finding and implementing the magic clue will be much harder than you might possibly expect. I gave up and designed around a headless approach since its the easiest fix.

You should try playing .wav files directly with the aplay command with -v to see what is happening. You may be getting underruns because the interrupt service is late. (Use Google to translate the geek mumbo jumbo).

 

[Ned Kelly]

Guys, I'm going to lock myself in my apartment for the next few weeks and try to get my spanking new Streamer II sorted with Linux. Even with Win7, there are periodic artefacts (not exactly clicks - more like a tiny 'skip' on a CD), and it seems more common when the Streamer is upsampling WAV to 24/96. We are talking something in the order of one skip every 3 or 4 songs, but it remains an annoyance.

 

I believe Demian is correct - I spend a lot of time in Resource Monitor watching processes grab their CPU slice/access the HDD etc and it doesnt take Einstein to deduce that this could be impacting the data transfer to the HRT DAC. Chris hasnt reported any such issues with his Atom-powered C.A.S.H server, but I still have things like SQL Server and a MySQL daemon running around on my netbook. Booting into Voyage-MPD will allow me to get back to basics, hopefully - will let you know how I get on.

 

 

[roscoeiii]

Have you tried contacting HRT about this issue? I found Kevin Halverson very helpful when I had probs with my HRT Pro. Very satisfied with the assistance I got from HRT. One of the things that makes them such a great company.

 

[Ned Kelly]

At about 3pm this afternoon, various issues that had plagued me with ALSA, file perms and other little annoyances finally went away, and mpd served up my music without anything getting in the way. The challenge now is a simple one - try to build my own headless server or simply accept that Demian has already invented the mousetrap I need. ATM, I'm completely over it and am going to spend a few days doing things that have little to do with either audio or Linux. I dont think I have ever looked forward to returning to work on Monday morning, but thats the frame of mind I find myself in tonight. As I said in my other thread, its masochism.

 

[1audio]

There is a strong feeling of accomplishment when it starts working. Some of the problems are quite vexing and these systems are very complex. I need to relearn the details of each subsystem every time I deal with it.

 

On the one hand there is also a feeling of control when you build it yourself, but sometimes its just too much. I once understood the details of every automotive subsystem. Today, if the car starts and runs I don't want to know more. I just want to drive.

 

The next steps for an MPD setup are the ones around controllers (clients), connecting to storage and optimizing the sound interface.

 

You are doing the right thing to disconnect and enjoy something else (a concert perhaps?)

 

 

[Ned Kelly]

Thanks for those words of support, Demian, but I dont even want to listen to music for the next couple of days - burnt out while burning in my Streamer II alongside the experimentation with mpd. The first law of audio should be something along the same lines as a lab experiment - you only change one variable at a time - but I put myself in a position where I just wanted to throw my netbook against the wall and go and buy a bloody iPod Nano. Unlike other DACs I have owned (admittedly with a retail price higher than the Streamer), there is a bit of a song-and-dance when you first start the HRT product up - manually unmuting it, selecting bitrate/samplerate etc and all of this has to be done from within the OS. When you don't know your way around amixer/alsamixer etc, its initially very frustrating - esp when I read about others who have simply plugged the thing in and it started working as soon as they typed 'mpc play' ! I think I have now memorized every line of mpd.conf, including the comments ....

 

Its more straightforward under Win7, but the Atom just isn't up to dealing with that footprint AND delivering skip-free audio, or at least it hasnt been for me. If nothing else, mpd confirmed that there is nothing wrong with my DAC - whether the issue could be resolved with buffer/cache settings is something I need to look at later in the week.

 

Having bought an asynch USB DAC, I now realise that it wont play with your server, but given that I don't own a balanced amp anyway, I guess its all a bit premature anyway. That leaves me with solutions like the Alix board - I'll take another look when my enthusiasm for tinkering returns.

 

Cheers,

 

Ned

 

[1audio]

Been there, done that, especially when the wrong setting "bricks" the box.

 

Specifics-

Alsa supports Async USB audio class 1 adding class 2 after 1.0.22 (maybe earlier).

MPD should work with USB audio directly

 

Here is an mpd.conf snippet that should just work for USB audio:

 

 

audio_output {

type "alsa"

name "USB"

device "plughw:1" # optional

mixer_device "hw:1" # optional

mixer_control "PCM" # optional

mixer_index "0" # optional

}

 

Paste it into the mpd.conf file amid the audio output statements.

 

Add this for volume control (I'm still unclear as to which really controls it, mpd seems to be changing this) (all you should need to do is remove the hash mark). If the DAC doesn't have internal volume control nothing will happen.

 

#

mixer_type "hardware"

#

 

If the dac isn't standards compliant you may have trouble. You should not need pulseaudio or any similar things. No asound.conf (remove it, its not used with this).

 

The other issue, and why we are preparing a specific system for USB audio, is that our platform, and many others, have a single USB host controller in the system. What this means is that all the USB traffic get routed through a single USB-PCI interface. Many chipsets have this on the southbridge chip so its not visible externally. The problem with this is if you have two high speed activities going on (USB drive and USB audio) you can run into contention and drop data. Linux may not be as highly tuned for this as other platforms. However adding a PCI-USB adapter to the motherboard alleviates this issue and will sound better. We have some interesting developments in process along these lines.

 

It also highlights something the USB audio crowd overlooks- all the USB traffic goes through the PCI interface, just like the direct connection sound cards. All of the PCI soundcards I know of are asynchronous to the buss and use DMA to get the data directly from RAM during playback. With async USB the same happens but the USB transactions are in the middle. The benefit of async USB is moving the sync point closer to the dac and ideally locking it to the dac. The PCI interfaces that support word clock locking (Lynx, RME, Mykerinos) do essentially the same thing as Async USB when used with compatible DAC's, however adding a lot of monkey-motion to the playback process as Chris will attest.

 

 

 

[Ned Kelly]

Thanks for all that info, Demian - its fantastic to get so much detail on computer audio in one place. My DAC doesnt have any volume control, but thats fine - I control that with the amp. If I had your mpd.conf stanza on Friday, I may have saved myself a lot of pain and suffering, but it has certainly been character building

 

 

 

[1audio]

I hope the tip helps.

 

The core of Linux success is skillful searching of Google. Knowing how to construct the search and evaluate the returns.

 

I wish Android was as simple for finding help. Its not as well organized but a similar learning experience.

 

[hifihere]

Hi Demian, I am intrigued by your mpd.conf settings. I don't mean to hijack this thread but your conf settings seem to apply to me.

 

My current project is to squeeze all I can out of Vortexbox. It uses MPD as a player and recently implemented USB 2.0 support.

 

Pulseaudio and ALSA are terrible sounding paths. I have been able to get large improvements for SPDIF output by setting hardware to "hw" or "spdif". This bypasses the volume control and improves sound tremendously.

 

What would be a good conf to get the most direct connection to USB?

 

 

 

[1audio]

Except by switching to OSS. Alsa is the framework for the sound card driver. It provides the hooks (API) for the player to pass the audio to the hardware. With mpd switching from plughw to hw with the latest version of mpd moves the process upstream. The code does exactly the same thing otherwise. The mpd code may be cleaner since it has fewer options to accommodate but the alsa code may be more closely synchronized to the buffers.

 

Pulseaudio has more in common with the windows or mac sound api and managing mixers and sample rate converters and a bunch of other stuff. Its designed for desktop audio and making a computer more automatic in that scenario (just like a Mac or PC). For a dedicated audio system it can be dispensed with since you don't need to play system sounds while other sounds are playing.

 

Dispensing with alsa or oss is only possible if the player software can talk directly to the hardware, which means the full driver code would need to be incorporated. (Shades of early dos programs.) It would also be difficult since Linux (posix, unix, apple OSX and even Windows NT+) all try to keep the hardware isolated from the application layer so a sick app does not kill the machine.

 

Using the "hw" does not bypass the mixer either. Its still functional if the hardware "enumerates" a mixer. You should not have the software mixer enabled (dig into the settings for both mpd and alsa) For devices that do not have a hardware mixer (usb to spdif for example) there will be nothing for the mixer to do.

 

I have tried this with the latest mpd and it all works as expected. I can't say that I hear a difference yet. The first round is to make sure it works. I also see no significant difference in cpu load. The change does simplify the settings a little. There are a lot of other under the hood changes in mpd that could have an effect on the sound if the software can have an effect.

 

[hifihere]

Thank you for your rapid response Demian.

 

I do realize that either ALSA or OSS are necessary with Vortexbox Player. I just spent 47 days experimenting with replacing ALSA with OSS4 with disastrous results so it's back to ALSA tuning for now.

 

I don't know what is specifically being bypassed by using "hw" but I can report that the software volume control is no longer functional and the muddy sound is transformed.

 

When USB is used the volume control is not functional either regardless of what hardware setting I enter so I assumed that USB was a more direct path (?)

 

I would like to remove Pulseaudio but that left Vortexbox crippled when I did.

 

When you said "I have tried this with the latest mpd..." were you referring to disabling the software mixer?

 

 

 

[wilbert-vb]

Hello there,

 

I'm a bit reluctant with responding in this forum after learning most from reading. I also don't pretend to be an audiophile expert.

I had this idea that OSS was outdated and replaced my the modern ALSA, but from researching I learned that ALSA is a reaction to the licensing changes of OSS years ago. OSS is up to date to modern hardware and available for several operating systems. ALSA is only available for linux.

 

Anyway, I tried to get OSS4 working on my linux system, and it recognized the USB DAC, but couldn't write data to it.

So I installed FreeBSD 8.2 on a test system and connected my USB DAC. That worked right away, without any issues.

 

So, I decided to reinstall my Atom based music server with FreeBSD 8.2-RELEASE. The new installation recognized my Music Streamer II right away and I had sound.

 

I found the setting to enable bitperfect sound:

Add this lines to /etc/sysctl.conf:

dev.pcm.1.bitperfect=1 and ofcourse disable auto_resample for musicpd (mpd on FreeBSD) in mpd.conf.

 

I noticed right away a significant difference with internet streaming, I listen to a classic NPR station and somehow I was misguided about bitrate/samplerate. I had to enable 44100:16:2 resampling in order to listen to this station.

With FreeBSD/OSS4 I had nothing special to configure to listen to the same station. And the Music Streamer II shows the LED 96K lit up.

The sound quality of this internet stream is so much better, unbelievable!

 

The local stored flac tracks sound so much better too.

Yesterday's linux/ALSA sounded dense and stuffed, today's FreeBSD/OS4 sound open and light. I hear the space around the performers, I hear the distance to the cough in the audience. I could not stop listening to my familiar CD's and just sit and enjoy. This is some serious upgrade in audio joy!

 

For me: farewell ALSA, welcome OSS.

 

[1audio]

I'm less committed to either but Alsa has more supported devices and I have it working well.

 

Its good to hear that OSS4 can work well.

 

Can you post any of the Alsa settings you were using so we can see what to avoid. From the description you are resampling everything to 96/24. The docs I saw for OSS4 resampling were sparse so I can't guess which resampler you were using. The default on Alsa is pretty marginal, but uses little CPU. The better ones use a lot more power.

 

[Miska]

Yesterday's linux/ALSA sounded dense and stuffed

 

Are you sure you weren't using what most distros have by default:

ALSA -> PulseAudio -> ALSA

 

Resamplers in ALSA and PulseAudio are fine for desktop "bling"-sounds but not for serious use.

 

 

[wilbert-vb]

I checked the LED on the USB DAC, while listening to 44.1/16 a track and you are correct, this gets resampled too. Hm, I thought that with dev.pcm.1.bitperfect=1 in /etc/sysctl.conf I was disabling any resampling. (And I checked that it's set correct with sysctl)

Now I'm confused, I need to research for that.

 

I didn't modify any ALSA settings, and I used hw:1,0 in mpd.conf.

The music server did not have pulseaudio installed It's a headless system without graphic desktop.

 

[wilbert-vb]

There was no pulseaudio installed on the headless server.

 

[hifihere]

I am glad to see this thread light up again after six months

 

Demian would you please provide specific details for a configuration of ALSA/MPD?

 

Earlier you suggested defeating the mixer for example so that and any other setting details would be appreciated.

 

My mission is to get a more transparent sound from the Vortexbox using ALSA/MPD. So far my biggest improvement in software settings was to move away from the Squeezebox control environment to direct control of MPD.

 

I expect a streamlined ALSA/MPD configuration would benefit many users . Perhaps this should be moved to a new thread where we could share our results.

 

[wilbert-vb]

I do agree with you, HifiHere,

 

I also started with Vortexbox, then decided that I wanted a lighter setup and re-installed with Debian/Squeezebox. And after some software issues I uninstalled squeezebox and installed mpd.

I thought that the sound quality improved, although I can't describe how.

 

And now I'm fiddling with FreeBSD/mpd, and I enjoy it!

 

The sad thing is, though, that on my iPod, MPoD is not as fancy as iPeng for remote control, especially with multiple cd-sets. And I miss Pandora internet radio.

(I get introduced to unknown albums by listening to this service)

 

Although my joy of listening to music does not depend on technical specs, I do like to understand the 'under the hood' part and look for improvements.

 

I see now the advantage of a tiny Alix or PogoPlug based system with USB-DAC that has access to a networked fileserver with the music tracks. Such setup makes it easy to re-install and 'play around'.

 

[1audio]

MPD settings are pretty hardware specific. I'm reluctant to offer too many suggestions and get people off in the wrong direction.

 

In the outputs section you can used settings like an alsa config file. If your USB hardware supports volume control you can configure that as well. Beyond the obvious like avoiding sample rate conversion if you can and talking directly to the hardware I can't offer much more advice without specifics and even with specifics different host hardware may require different settings.

 

I'm really doing all I can manage to support the various Auraliti platforms and most of the software details have been published. I'm doing hardware tweaks now, there is much more to be gained in hardware improvements at this level.

 

 

 

[hifihere]

Wilbert-vb I absolutely agree with you regarding iPeng. Recently I made similar comments on the VB forum. It is the most polished controller I have used. I left MPod quickly and moved to MPDroid. I like the new version that came out last week but still miss the slick features of iPeng.

 

My main reason for leaving the Squeeze platform was the lack of gapless playback which I consider essential. An added surprise was how much better the sound was when I moved to MPD control with MPod and MPDroid.

 

Demian, maybe I have done all I can with the config at this point. I agree with your choice of the Juli@ for digital output. It is the best I have heard so far for that use.