For hardware for audio playback I think there are several roads ahead, all valid.
First and my current goal, is a simple package that will play the original audio stream untouched, no level adjustments, no sample rate conversions noting but the raw data stream at its original sample rate. For the most sophisticated DACs (like the Berkeley Audio or the Pacific Microsonics) anything upstream may and probably will reduce the performance of the DAC. Getting a low jitter stream at the native sample rate is paramount. And with HRx and other high resolution audio formats becoming available a means for playing them is in demand.
Second would be a system that supports eq/crossovers and any other similar types of audio processing. This gets harder fast and needs more horsepower. I would like a system that would not change sample rates but rather switch co-efficients to match the sample rate of the source.
There are a raft of other issues around user interface and distributed audio and network/internet sources as well. And no single solution will handle all of this well.
First and my current goal, is a simple package that will play the original audio stream untouched, no level adjustments, no sample rate conversions noting but the raw data stream at its original sample rate. For the most sophisticated DACs (like the Berkeley Audio or the Pacific Microsonics) anything upstream may and probably will reduce the performance of the DAC. Getting a low jitter stream at the native sample rate is paramount. And with HRx and other high resolution audio formats becoming available a means for playing them is in demand.
Second would be a system that supports eq/crossovers and any other similar types of audio processing. This gets harder fast and needs more horsepower. I would like a system that would not change sample rates but rather switch co-efficients to match the sample rate of the source.
There are a raft of other issues around user interface and distributed audio and network/internet sources as well. And no single solution will handle all of this well.
1audio said:
Since you stuck your head up, can I ask you to try a test.
Please try foobar with the ABX plugin. Use Sox (or some other resampler which scores well in the link above) to resample your most revealing audio track, then use headphones for best transparency (sure use your hifi system if you prefer).
Foobar should be able to give you sample perfect output at a variety of output rates
I would be very interested if you can detect a difference...
If you can't hear a difference of course then you potentially free up a bunch of parameters making it simpler to pursue your goal... Let us know - honest it's quite an interesting experience.
Try it with some other things like converting to MP3 at various bitrates - it's amazingly how many "night and day differences" melt away when you blind test them. Everyone should try it...
I'll be happy to explore per your suggestions but it will take a few days. However a little clarity here would help. Are you suggesting using SOX to upsample from 44.1 or downsample from a higher sample rate? I would first explore downsampling from 176.4 material to see if it degrades audibly (it always has in the past). I would use the SRC offline for this. I don't have any MP3 material myself but I suppose I could encode and decode for what its worth.
I use MPD and it is configured to automatically switch sample rates through to the output as I described with full 24 bit resolution throughout. Only in the current git for now.
I'll check out the ABX feature of foobar but I have always found the abx paradigm to make the whole process unpleasant and harder to tell differences. Perhaps its better as a software plugin than some clunky hardware.
I use MPD and it is configured to automatically switch sample rates through to the output as I described with full 24 bit resolution throughout. Only in the current git for now.
I'll check out the ABX feature of foobar but I have always found the abx paradigm to make the whole process unpleasant and harder to tell differences. Perhaps its better as a software plugin than some clunky hardware.
1audio said:
Well, to be honest, I would actually not expect you to be able to tell the difference between stuff sampled up or down, as long as it's done well.
I had in mind that you probably had 44Khz audio and you could set it to some multiple upwards, however, either way would be instructive (where did you find some 176Khz material?)
As for the MP3, obviously you need to generate it yourself or else you won't have the high quality original to compare with!
Use a decent encoder and see at what point you can reliably tell the difference. It's quite likely to be quite a low bitrate actually unless you pick some type of music that the encoder doesn't handle well. I found this quite an interesting test actually...
1audio said:
Well, you struck the nail on the head really. Obviously you don't want the process to be a pain, hence try ABX and foobar - I found it fairly simple and convenient. However, you are right, once you try to prove the difference between two samples I personally find that many of those "obvious differences" can shrink away and be quite hard to prove.
Obviously if you find that your resampler is inaudible in practice then you potentially opened a bunch of avenues for yourself
Re: Converting Ape on Linux (with GUI tool)
well, thanks. But IMHO it's just so easy to simply use "mac" (the monkeys-audio codec) to convert .ape back to wav or "shntool" to convert ape to just about anything else directly from the cmd line...
Otherwise, if you can't live without a GUI there are plenty of other applications which are more specific to sound file conversion than k3b. For KDE a very versatile one is SoundKonverter.
It supports:
Decoding: ogg, mp3, mp2, m4a/mp4, aac, 3gp, mpc/mp+, flac, ape, wma, asf/asx, ra, rv, rm, avi, mpeg, wmv, qt/mov, flv, ac3, au/snd, shn, tta, bonk, ofr, ofs, wv, la, pac, spx, mid, mod/s3m/stm/ult/uni/xm/m15/mtm/669/it, wav
Encoding: ogg, mp3, mp2, m4a, aac, mpc, flac, ape, ra, ac3, au, shn, tta, bonk, ofr, ofs, wv, la, pac, spx, wav
Is it enough?
P.S.: sorry for the late answer...
well, thanks. But IMHO it's just so easy to simply use "mac" (the monkeys-audio codec) to convert .ape back to wav or "shntool" to convert ape to just about anything else directly from the cmd line...
Otherwise, if you can't live without a GUI there are plenty of other applications which are more specific to sound file conversion than k3b. For KDE a very versatile one is SoundKonverter.
It supports:
Decoding: ogg, mp3, mp2, m4a/mp4, aac, 3gp, mpc/mp+, flac, ape, wma, asf/asx, ra, rv, rm, avi, mpeg, wmv, qt/mov, flv, ac3, au/snd, shn, tta, bonk, ofr, ofs, wv, la, pac, spx, mid, mod/s3m/stm/ult/uni/xm/m15/mtm/669/it, wav
Encoding: ogg, mp3, mp2, m4a, aac, mpc, flac, ape, ra, ac3, au, shn, tta, bonk, ofr, ofs, wv, la, pac, spx, wav
Is it enough?
P.S.: sorry for the late answer...
done that. ( Though not using foobar... I don't do windows, ever!
). I can tell mp3 apart from lossless even for LAME "extreme" VBR and 320Kbps CBR settings.
For lower quality mp3s, on my system I can promptly detect 'em even without comparing.
If you carelessly listen to a (high quality) mp3 alone it may seems good. But - given you have a good enough audio system (that's NOT just some random computer sound card with some random hearphones), if you can compare it with the lossless original the difference is obvious.
The lossy codecs simply "wash out" the music. At a first glance sometimes the good mp3s may even seem to sound "better" then the original: they usually sounds somewhat "cleaner". Then you realize that they seems cleaner because they lack sooo many little details, ambience, etc.
Problem is, so many audio system (likely all of the "consumer" types, but also way too many supposedly "hi-end" ones) sounds like mp3 themselves.
I guess that's why so many people can't tell the difference between mp3 and "real" music.
Actually, sometimes I use the "mp3 test" to check the quality of an audio system: if I can't clearly hear the difference between a "good mp3" and the real thing (lossless file), than the system isn't any good... no matter how expensive it may be. Using mp3s of different quality, you may even sort of "quantify" the quality of the reproduction system!
Hi folks.
Good news for folks running the .wav format and MPD/Minion.
I just tested the newest version of Minion (Firefox-MPD add-on) : 1.99.23
Chris - the guy in charge for Minion introduced .wav tagging from filenames (tag-guessing). I am not aware of any other application under Linux doing this.
( You tell me )
On my file-name-format: "Number-Artist-Album-TrackTitle" it works quite well.
For the time being you need to configure a special key in Firefox to enable it.
1. Type "about:config" in the address field .
2. Right click in the main window-field- select: "new""boolean"
3. add "extensions.mpm.guess_tags"
4. set to true
5. Restart FF
That's about it.
I am still discussing with him to get rid of underscores in tags ( Standard feature supplied by EAC). I got all blanks with underscores replaced (for better Unix handling) on my filenames.
Cheers
Good news for folks running the .wav format and MPD/Minion.
I just tested the newest version of Minion (Firefox-MPD add-on) : 1.99.23
Chris - the guy in charge for Minion introduced .wav tagging from filenames (tag-guessing). I am not aware of any other application under Linux doing this.
( You tell me
)On my file-name-format: "Number-Artist-Album-TrackTitle" it works quite well.
For the time being you need to configure a special key in Firefox to enable it.
1. Type "about:config" in the address field .
2. Right click in the main window-field- select: "new""boolean"
3. add "extensions.mpm.guess_tags"
4. set to true
5. Restart FF
That's about it.
I am still discussing with him to get rid of underscores in tags ( Standard feature supplied by EAC). I got all blanks with underscores replaced (for better Unix handling) on my filenames.
Cheers
@ewildgoose
Did you try MPD+Minion+brutefir as I describe it in the Wiki?
http://picasaweb.google.com/lh/photo/FDp-kMbCFN4Xge6UceQ6nQ?feat=directlink
Did you try MPD+Minion+brutefir as I describe it in the Wiki?
http://picasaweb.google.com/lh/photo/FDp-kMbCFN4Xge6UceQ6nQ?feat=directlink
ewildgoose said:
The great thing with MPD and "pipe-out":
Every system can be differently configured, with a great set of options. You can control any audio-client from any place in the network. It can be controlled from any place in the network, even your mobile phones. Pretty much like the Sonos system but much more flexible, because you can use any HW.
If you ask me. I don't need a different application for everyday-use.
Hi folks.
1.
I still have an outstanding issue, with Onvinyl.
He doesn't get the internet .pls streams going within MPD
I added some more radio stream examples to the MPD Wiki.
You might want to try these.
2. Explanation to the new Minion WAV-tag-guess
The logic behind the new .wav-tag guess mechanism in Minion is following:
Minion is looking for "fields", when examining the filename:
Field seperator is fixed: "-" (dash)
Field1-Field2-Field3.....X,Field_final.wav
Scheme:
Field 1 =Tracknumber
Field 2 = Artist
Field 3....X = Album
Field_Final =Track-Title
Everything what is between Artist and TrackTitle is put in the Album field.
Since the most common filename scheme is <track-number><artist><album><track-title>
above will work pretty reliable.
I just went through my collection ( and cleaned all my mess up).
I guess I had a 95% hitrate (on my already well structured filenames)
Biggest problem are usually the VA (sampler) and Classical and all these dashes in names.
This filename business takes me more time than the ripping itself.
Chris will soon launch Minion 1.99.24, which corrects some small problem around this. He'll also remove the "underscores" of filenames.
I hope I don't bore you to death with all this.
Cheers
1.
I still have an outstanding issue, with Onvinyl.
He doesn't get the internet .pls streams going within MPD
I added some more radio stream examples to the MPD Wiki.
You might want to try these.
2. Explanation to the new Minion WAV-tag-guess
The logic behind the new .wav-tag guess mechanism in Minion is following:
Minion is looking for "fields", when examining the filename:
Field seperator is fixed: "-" (dash)
Field1-Field2-Field3.....X,Field_final.wav
Scheme:
Field 1 =Tracknumber
Field 2 = Artist
Field 3....X = Album
Field_Final =Track-Title
Everything what is between Artist and TrackTitle is put in the Album field.
Since the most common filename scheme is <track-number><artist><album><track-title>
above will work pretty reliable.
I just went through my collection ( and cleaned all my mess up).
I guess I had a 95% hitrate (on my already well structured filenames)
Biggest problem are usually the VA (sampler) and Classical and all these dashes in names.
This filename business takes me more time than the ripping itself.
Chris will soon launch Minion 1.99.24, which corrects some small problem around this. He'll also remove the "underscores" of filenames.
I hope I don't bore you to death with all this.
Cheers
Hi folks.
The new Minion 1.99.24 is available for download.
Have a look at my current setup incl. wav-tag-guessing.
My Minion setup incl. .wav-tags from filenames
It is IMO a great feature.
I am still in discussion with Chris on certain improvements.
Even though it is a preliminary version I highly recommend this application in its current status.
Cheers
The new Minion 1.99.24 is available for download.
Have a look at my current setup incl. wav-tag-guessing.
My Minion setup incl. .wav-tags from filenames
It is IMO a great feature.
I am still in discussion with Chris on certain improvements.
Even though it is a preliminary version I highly recommend this application in its current status.
Cheers
Hi folks.
Below issue I'd call a bit off-topic and belongs in the area "UNIX System administration".
I think it is really important and more than worth do bring it up for discussion once. (If people are interested to discuss this at all)
I spent quite some time recently to get my collection straight (filenames/albumarts/new rips).
Counting the overall hours I spent to get the entire collection on the HDD, at least the same time to get the filenames in a standardized format and not to forget the exhausting manual download of albumarts - suddenly one thing hit me hard:
"What about BACKUP?" I felt a slight dizziness/stress slowly crawling up. "What if i'd loose all that stuff. No - don't think about it!"
I have to admit that backups were not really the first priority on my agenda.
Here and there I installed a new Linux, usually before a HDD crashed and copied all the data over there. This kept me going. Lucky me!
I also used to copy stuff to my NAS, which took lightyears to get the data on and off this device.
In the end I gave up on that one and just a third of my collection were stored on the NAS.
The rest were handled without a backup. Shame on me!
Anyhow, after above question popped up this strange feeling wouldn't disappear -- 4 hours later I had an external 1TB HDD on my desk. No - I couldn't wait for an online-order shipment. I'd call this situation a slight panic situation.
Then the fun part started: What to do? and How to do it?
Just "cp". NO - that won't work. A ready-made-blackbox-tool: NO - that won't do either.
I usually want to know what's going on (not as deep as phofman of course , but that deep that I think I'd have control over things) .
However - The whole subject is not as easy as it might seems - if you want to do it in a rather "professional" well structured and organized manner.
If you surf around a bit you'll realize that the whole thing is a real challenging exercise, if you want to do it right.
To make long story short rsync and dd ( I won't go deeper on these tools now) are the obvious options I've chosen.
Of course there is more to it then just choosing the two commands, formatting the disk, choosing the right filesystem, handling the master boot record, the partition configuration, you might want to think about incremental backups, and backups over network, cloning disks, cleaning before dumping etc,etc,...... Further you need to structure your partitions, filesystems and names properly. And don't believe a /dev/sdb HDD is always a /dev/sdb. You pull out /dev/sda and suddenly sdb becomes sda after the next reboot.
(I followed an advise to remove my UUIDs (fstab/grub) and to get back to plain device ids - don't do that! This can cause severe trouble, if you use mutiple disk configurations), Then you find out that MS creates partitions in a rather strange way - not ending on cylinder boundaries - which might cause trouble on the disk and so on and so on.
I ended up reconfiguring my entire system and writing a script covering most of the features I was looking for .
I even tested a "Restore" already.
Very stressful exercise.
(Perhaps some of you know this situation: You type in a command. And then you're afraid to hit "return". Because you know, if you have done something wrong, you can mess up your system. It'll take a minute or two. You do check all things the fifth time and then you hit "return" and close your eyes. No,No - you don't want to to see what's happening.
Usually these situations happens ones in while. Last week I had every day 5 of them. This are adrenalin-shots at its best. Who the hell needs drugs?
I learned quite a lot of stuff during the last week.
I feel much better now!
How do you guys feel? Bad conscious?
Perhaps you let us know how you are running your backups?
Perhaps we all can learn a bit more about the subject!
Cheers
Below issue I'd call a bit off-topic and belongs in the area "UNIX System administration".
I think it is really important and more than worth do bring it up for discussion once. (If people are interested to discuss this at all)
I spent quite some time recently to get my collection straight (filenames/albumarts/new rips).
Counting the overall hours I spent to get the entire collection on the HDD, at least the same time to get the filenames in a standardized format and not to forget the exhausting manual download of albumarts - suddenly one thing hit me hard:
"What about BACKUP?" I felt a slight dizziness/stress slowly crawling up. "What if i'd loose all that stuff. No - don't think about it!"
I have to admit that backups were not really the first priority on my agenda.
Here and there I installed a new Linux, usually before a HDD crashed and copied all the data over there. This kept me going. Lucky me!
I also used to copy stuff to my NAS, which took lightyears to get the data on and off this device.
In the end I gave up on that one and just a third of my collection were stored on the NAS.
The rest were handled without a backup. Shame on me!
Anyhow, after above question popped up this strange feeling wouldn't disappear -- 4 hours later I had an external 1TB HDD on my desk. No - I couldn't wait for an online-order shipment. I'd call this situation a slight panic situation.
Then the fun part started: What to do? and How to do it?
Just "cp". NO - that won't work. A ready-made-blackbox-tool: NO - that won't do either.
I usually want to know what's going on (not as deep as phofman of course
, but that deep that I think I'd have control over things) . However - The whole subject is not as easy as it might seems - if you want to do it in a rather "professional" well structured and organized manner.
If you surf around a bit you'll realize that the whole thing is a real challenging exercise, if you want to do it right.
To make long story short rsync and dd ( I won't go deeper on these tools now) are the obvious options I've chosen.
Of course there is more to it then just choosing the two commands, formatting the disk, choosing the right filesystem, handling the master boot record, the partition configuration, you might want to think about incremental backups, and backups over network, cloning disks, cleaning before dumping
etc,etc,...... Further you need to structure your partitions, filesystems and names properly. And don't believe a /dev/sdb HDD is always a /dev/sdb. You pull out /dev/sda and suddenly sdb becomes sda after the next reboot.(I followed an advise to remove my UUIDs (fstab/grub) and to get back to plain device ids - don't do that! This can cause severe trouble, if you use mutiple disk configurations), Then you find out that MS creates partitions in a rather strange way - not ending on cylinder boundaries - which might cause trouble on the disk and so on and so on.
I ended up reconfiguring my entire system and writing a script covering most of the features I was looking for .
I even tested a "Restore" already.
Very stressful exercise. (Perhaps some of you know this situation: You type in a command. And then you're afraid to hit "return". Because you know, if you have done something wrong, you can mess up your system. It'll take a minute or two. You do check all things the fifth time and then you hit "return" and close your eyes. No,No - you don't want to to see what's happening.
Usually these situations happens ones in while. Last week I had every day 5 of them. This are adrenalin-shots at its best. Who the hell needs drugs?
I learned quite a lot of stuff during the last week.
I feel much better now!
How do you guys feel? Bad conscious?
Perhaps you let us know how you are running your backups?
Perhaps we all can learn a bit more about the subject!
Cheers
It's not that well documented, but unlike with windows, udev on linux allows you to have persistent names for most hardware including USB sticks, network cards, removable hard disks, etc
This means that your wireless USB stick will be eth2 each time you insert it and always that same eth2 device. unplug it and plug in another one and it becomes eth3, etc. (this catches people out more often than it's helpful I guess...)
So with a bit of understanding you can give your removable hard drive a unique mount location and check it's mounted before backing up
With some fairly simple udev magic you can even achieve things like running the backup as soon as you connect your removable drive. This is fun for things like memory sticks where you can basically script an operation to dump off all your photos (say into date named folders) simply because you inserted the card into the machine.
rsync is probably fine for what you have in mind, however, remember you have --dry-run to show you what will happen first
Good luck
Ed W
P.S. I would agree that the cheapest backup remains buying some 1TB drives....
This means that your wireless USB stick will be eth2 each time you insert it and always that same eth2 device. unplug it and plug in another one and it becomes eth3, etc. (this catches people out more often than it's helpful I guess...)
So with a bit of understanding you can give your removable hard drive a unique mount location and check it's mounted before backing up
With some fairly simple udev magic you can even achieve things like running the backup as soon as you connect your removable drive. This is fun for things like memory sticks where you can basically script an operation to dump off all your photos (say into date named folders) simply because you inserted the card into the machine.
rsync is probably fine for what you have in mind, however, remember you have --dry-run to show you what will happen first
Good luck
Ed W
P.S. I would agree that the cheapest backup remains buying some 1TB drives....
Backup is the hidden challenge for all digital entertainment media. I know people with 15-20 TB of movies. All on a large raid. I point out that RAID is not BACKUP!!! A single disk failure can be recovered but a multi disk crash is total, as is a controller failure (except for a Cable Company no one needs RAID throughput to watch movies, let alone music) so I push software raid and a full paralleled disk backup. And test the restore process (I need to do that as well) because a defective backup is worse than useless.
A larger record collection was worth far more that the playback hardware, even pretty premium playback hardware. The same for a CD collection of any size. In the new download, copy to disk age the original may not be on the shelf and one crash can wipe out 1,000's of recordings and as Soundcheck realized 1,000's of hours curating a collection (that should not need it if Metadata didn't just stink).
I would pay/contribute to a good reliable backup/restore/manage entertainment media project. I bet there are a lot of others who would as well. Something like FreeNAS is a foundation but not the right solution.
A larger record collection was worth far more that the playback hardware, even pretty premium playback hardware. The same for a CD collection of any size. In the new download, copy to disk age the original may not be on the shelf and one crash can wipe out 1,000's of recordings and as Soundcheck realized 1,000's of hours curating a collection (that should not need it if Metadata didn't just stink).
I would pay/contribute to a good reliable backup/restore/manage entertainment media project. I bet there are a lot of others who would as well. Something like FreeNAS is a foundation but not the right solution.
At home I copy using cp/rsync to an external eSATA HDD. My legacy windows partition is in vmware, hence cp/rsync are sufficient. I do not need backup history at home, that makes things pretty simple. Who cares about downtime for offline backup at home.
At work we need automated backup with history (several days, weeks, months, years). Backuppc http://backuppc.sourceforge.net/ has been serving us well for years. We use the history pretty often. Since the file structure uses hardlinks, the used space grows surprisingly slowly over time. Backup of the backup server to offline drives is provided by SW Raid1 - mirror. The background raid synchronization gives us almost no downtime of backuppc.
For external drives I can recommend eSATA/USB HDD docks http://www.welland.com.tw/html/docking/docking.html. Inexpensive (15EUR ?) 2 x eSATA PCI-e controller with hot-plug support hits 200MB/s synchronization speed to two striped (RAID0) Samsungs 750GB in the docks.
At work we need automated backup with history (several days, weeks, months, years). Backuppc http://backuppc.sourceforge.net/ has been serving us well for years. We use the history pretty often. Since the file structure uses hardlinks, the used space grows surprisingly slowly over time. Backup of the backup server to offline drives is provided by SW Raid1 - mirror. The background raid synchronization gives us almost no downtime of backuppc.
For external drives I can recommend eSATA/USB HDD docks http://www.welland.com.tw/html/docking/docking.html. Inexpensive (15EUR ?) 2 x eSATA PCI-e controller with hot-plug support hits 200MB/s synchronization speed to two striped (RAID0) Samsungs 750GB in the docks.
Thought this was the best place to ask
Thought this was the best place to ask, any help with this from linux audio buffs...?
http://www.diyaudio.com/forums/showthread.php?s=&threadid=142322
Thought this was the best place to ask, any help with this from linux audio buffs...?
http://www.diyaudio.com/forums/showthread.php?s=&threadid=142322
soundcheck said:Foobar/Asio/Windows... You're at the wrong place.
My preference by looking at what you'd like to achieve:
MPD piped into brutefir - No need for jack or netjack.
Good luck
I'm close to agreeing... just clinging on to the flexibility of being able to play anything I like on my windows machine and have it processed properly by the linux box and fed over the mains...
I run a meagre dsp machine (duron 1.6ghz etc), and would like anything I have on my main pc to be playable.... so my next port of call is probably mpd running on the linux box with a permanent share to my music directory over the network to my windows box, controlled through minion as your excellent guide suggests...
Logically, its probably the better solution as well... Rather than have a foobar asio output synced with the clock of the dsp machines card over the network, stream the raw data via a share over the network and then decode it on the linux box... takes another link in the chain out....
Still if anyone has got the jackrouter asio plugin working it would be a relief to do so myself, if only for self satisfaction....
Remotely-controlled MPD is an alternative to netjack hooked to sound system infrastructure (asio) only for pure playback. What we need is a way to hook all asio applications (e.g. DVD players) to linux DSP machines with the actual sound card. I understand the requirement, since my little router-as-network-soundcard project has the same need - ability to hook to linux as well as windows.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.