On the eve of Mandrake’s latest release, Austin Acton has published a guide at DesktopLinux.com to create a digital audio workstation using Mandrake 9.1. This ‘HowTo’ describes how to set up a professional quality audio workstation in an afternoon or less without compiling or text editing.
The Mandrake Audio Workstation HowTo
Submitted by Jill 2003-03-22 Mandriva, Mandrake, Lycoris 27 Comments
I think Pro-Audio on Linux is a catch-22 situation. No software until you have lots of users. No users until you have software.
A distribution is needed where you can just put a new PCI card in, switch on, “Have detected new device, please insert driver CD”, voila, hardware and respective driver/software installed and configured automatically. (Windows/MacOS anyone?)
GNU/Linux distributions will never get there IMHO.
Maybe it’s time we had the linux kernel but without the UNIX type stuff. What I mean is use the kernel but that’s it. Have the rest like win/mac/beos. One standard platform for Linux Audio. Maybe something like the Blue Eyed OS is trying to achieve?
As programs like ardour (ardour.sf.net) mature, it will be a relatively simple task for a distribution dedicated to linux audio to emerge that can detect sound cards and use ALSA, jackit, etc., along with the right kernel patches.
Ardour has almost reached a feature level that competes with the big guns, soon it’ll just have to squash some bugs and then it’ll be ready for prime time.
Pro audio linux will be realized in the not-so-distant future, and having UNIX as a base is not the problem. Actually, OSX (built on UNIX) makes a far better recording studio than Windows, who’s architecture require a really fast, heavily loaded system to do pro-audio.
Distributions can already auto-detect hardware (even after installation). What about UNIX prevents that from happening?
It is already configured, but I never hear anything about it. It sounds great for audio/video/desktop use. I haven’t tried it but I hope to find time soon.
“Pro audio linux will be realized in the not-so-distant future, and having UNIX as a base is not the problem. Actually, OSX (built on UNIX) makes a far better recording studio than Windows, who’s architecture require a really fast, heavily loaded system to do pro-audio”
Well, I ‘d like to hear why. Windows, since windows2000, is a very good OS for pro audio. Sofwares like Nuendo are afaik very good ( and yes, ardour, even if the project is very insteresting, is far, far, far away from products like nuendo ). I think the reputation of apple is a little over estimate for music. mac were a lot better 5-6 years ago, but now… The only things windows lack is pro tools ( but nuendo or samplitude will harm digidesign a lot, I think ), and Digital perfourmer.
The number 1 pb for linux is : installation. I don’t personnaly think linux is very difficult to install, bu I know what a hard drive is, what a partition is, what a driver is, etc… Most people find windows ( and even mac ) too bloated, too difficult to use. GNU/Linux needs a very easy install, like BeOS, winXP or QNX ( I have never installed Mac OS, so I can not say anything about that ).
There is also the lack of drivers ( an exception RME cards, who are very good, and have good audio drivers under linux ). Maybe the coming of alsa in 2.6 release kernel will change that. There is also the problem of the very bas scheduler in 2.4 kernel, but I alsao heard that it will change in the 2.6 ( I haven’t played yet with 2.5.* kernel to see the changes… )
The last problem is the lack of standart for audio plugin : the present one is not very good, isn’t verty practical for automation.
i think that alsa doesn’t support the middle-to-high end audio devices. what about echoaudio, creamware, digidesign, motu?
an audio workstation for a dj who plays with trackers and stereo editing on soundblasters and yamaha cards is completely useless. First thing to achieve should be a good sopport for the pro-cards and a plugin protocol like the VST or the mac osx audio subsistem
Seems like the current state of Linux audio apps is like many other things – ‘good enough’ for some people, but not quite ready to take on the big boys.
However, it’s nice to see that they’re making inroads. I wish them the best of luck
I just installed Mandrake 9.1rc2 and found that my drakconfig tools are all missing their words. When I run #urpmi.setup I get a window with a series of blank buttons. Does anyone know of an easy fix?
Linux audio is very interesting at the moment, there are already a number of powerful and mature apps like PD, Jmax, Freqtweak, Lilypond, Audacity, Sweep and Csound. I was surprised not to see a mention of some of these in the article, I guess he only covered apps that come with Mandrake. Sweep for instance is sponsored in by Pixar and CSIRO, and does a few things no other editor can….
Ardour and Jack will provide the last missing links to provide a complete DAW.
The LADSPA plugin standard has become widely accepted due to it’s simple API and solid automation capabilities. It’s worth checking out http://www.plugin.org.uk to get a good selection of them.
The inclusion of ALSA in the 2.6 kernel will NOT make more sound cards magically work with Linux, you can download ALSA 0.9.1 now, and use it with almost any 2.4 kernel to get the same ALSA support as you would with a 2.6 kernel.
At the moment you can do a lot with Linux Audio. I run a Linux box in conjunction with my Win2k DAW to access all the synthesis and processing tools that on Windows remain a few versions behind, are not as stable, or are not available at all. There really is a vast amount of Linux audio software out there, a lot of it is admittedly quite academic, but for those who want to experiment, it’s all there for the taking.
I am working at the moment on a music oriented version of Knoppix, as it’s very easy to customise, and has great hardware detection. Another bootable media oriented distro to try is dynebolic http://dynebolic.org/
Steinberg have versions of Nuendo and Cubase running on Linux already, but are in the ‘wait and see’ stage.
Jack enables you do do a few cool things that are available only in very restricted forms like Steinbergs ReWire on other platforms, as it’s an open standard. You can, for instance use an insert on a mixer channel in Ardour to connect to seperate stand alone apps, and connect those apps together independently, and have the result appear on an insert return on another mixer channel in Ardour. I don’t think this is possible on ANY other platform.
Most distributions havent quite gotten there yet, but I’d like to mention that the installations for xandros and lindows and easy enough to be done by complete and total newbies to computers in general, not just GNU/Linux. For device detection and setup, lindows is ‘ok’, and many others dists are good, but xandros is damned amazing in this regard, you dont even have to THINK about drivers or need driver disks. Want examples?
Installed xandros on a machine with a voodoo3. Shut off machine without shutting down, replaced video card with a matrox, booted… Xandros presents a login prompt like nothing happened. Popped in a sound card. Turned on machine. Logged in. Got the kde login sound without having to touch any setup whatsoever. I added a tv tuner, booted, installed motv, and watched tv. the only setup was setting up motv. installed cd burner, burned some cds…. i think you get the picture here and yes i did it all one by one to test it out and see if it was really worth it and i didnt get a cd burner till much later.
Installation and setup issues are negligible and no longer fit in this kind of discussion!!!
As for the quality of current SOUND apps…. Others might have a different opinion but I agree they arent even close to being ‘there’ yet. It’s not even at the serious hobbyist level yet, nevermind professional… At least for electronic musicians. Unless you happen to really like impulse tracker clones and fugly interfaces to half-written software, of course. For multitrack stuff and other functions, there is satisfactory software, but not in the areas I personally like and want to use. I want to see a linux rubber duck, fruity loops, misc propellerhead software, dreamstation… MMMmMMMmm yeah… that’d be nice.
For now, until the programs mature, I can use:
Audio -> Ardour, and
MIDI and Audio Sequencing -> Rosegarden
Think of it as replacements for Cool Edit Pro 2, and Cakewalk Sonar 2.
For example, I have one system with a Celeron 750, 256 MB PC100 RAM, and a 5000 RPM drive. With it, I can record three or four mono tracks in full duplex, but no more
This concerns me. Why is so much power needed? My old DAW, Cakewalk, Celeron-400, 128Mb under Win98 could record 16 audio tracks at once, and mix down 32 tracks to 2 without any problems.
Is the author over cautious or does Linux really need that much iron to get the audio performance of an Atari ST circa 1986?
“It’s not even at the serious hobbyist level yet, nevermind professional… At least for electronic musicians.”
Ever heard of Jmax? Well, if you consider IRCAM not to be even serious hobbyists, then you have a little to learn about electronic music.
Heard of Csound? I guess MIT can’t be taken seriously either.
Heard of Pure Data? SuperCollider? These are all some of the most powerful synthesis programs in the world, and there are many more.
There was a Rebirth clone for Linux once, called ‘Reborn’. It did not contain any of the propellerheads code, and sounded identical, or perhaps a little better than the original. You could even load mods from the original Rebirth. Sadly, it was pulled after a complaint from Propellerheads.
I agree about the hardware installation though, and that there is not much good instant gratification software like rubber duck.
It does pay to throw your net a little wider when experimenting with sound…. Here is my fave page.
A great number of issues get in the way of linux audio. A major issue is the variable quality of audio device drivers, tho alsa drivers tend to be quite consistant, fast, and featurefull compared to free and commercial oss drivers in my experience.
Another major issue is latency, the time between an interupt and a wakeup. Latencies can be in the hundreds if milliseconds or more, causing all sorts of nasty things like dropped audio, noticable delays between sequencing information being received and processed, pops and noise, and a large number of other possibilities. Thanks to the low latency patch you can reliably get latencies down to under a few milliseconds most of the time and almost never more than 200 in the worst-case scenario. combine low latency and the kernel pre-empt patches and the worst case is more realisticly under 10ms. these are both unsupported patches for 2.4, pre-empt has been integrated into the 2.5 development kernel.
Some people complain of being unable to play mp3s on a 1Ghz machine or similar such activities. This is actually possible, and is most likely because most of the time DMA is not enabled by default. Install hdparm and enable DMA on your drives and this problem will disappear. INSTANTLY. This is probably the source of many complaints about linux usability. I’ve seen my box crash and burn under normal to semi-heavy use with dma off. I dont know of any dists off-hand that try to enable dma for drives, but i’m probably just not paying attention.
As an aside, compiling highly optimised binaries for ones’ machine can do wonders for performance. Check out gentoo. Its; beautifull. The gentoo kernel includes pre-empt support if you enable it. I wish there was a more elegant solution like mplayer’s runtime cpu detection available in a reasonable way for optimising all binaries, without the overhead of every app doing cpu detection.
Linux wasnt designed with realtime applications and audio content creation in mind, but the situation is improving rapidly and I have a very positive outlook for the future. Go team. *waves pompoms* At the moment it’s pretty damned far from the mark but it’s getting there pretty quckly. I want my toys now damnit.
“Ever heard of Jmax? Well, if you consider IRCAM not to be even serious hobbyists, then you have a little to learn about electronic music.”
Sure, but most DAW are not used to realise “Derive I et II”, they are used to record a rock band or a singer and a guitar or urban music or a string quartet.
In these areas the current s/w available on Linux would attract very very few Mac or Win based audio professionals.
csound is a programming language for sound, and not quite what I had in mind. It doesnt quite fit into my idea of an environment for creating music where an inspiration must be played with and recorded while it’s still there.
jMAX looks interesting, but isnt it written in…. java? It still doesnt quite fit.
I like your term instant gratification software, and I guess that’s where my interest lies. And as for the cool software out there, it’s almost all in alpha/beta and has a disgustingly ugly interface, if any at all.
The fact that reborn existed at doesnt anymore is kinda depressing. :/ I also think our views on what makes the cut are drasticly different. And since they are opinions, it wouldn’t make much sense defending them as fact, so I wont. That’d be just plain silly of me since for you things are obviously working well. ;p
“A major issue is the variable quality of audio device drivers”
An even bigger issue is the complete absence of audio device drivers for nearly all pro-audio equipment.
Right now the choice for a pro multi-channel audio I/O card under Linux is RME or RME. There are probably 40 or more that support MacOS-X or Win-XP. Its the same for other audio hardware.
The reason is not quite the chicken and egg situation (more users/more support) many make out.
Manufacturers complain about changing APIs and competing APIs, manufacturers are unconvinced about the GPL and its viral implications for their intellectual property. Closed source driver releases are distained by the Linux community. Licensing between manufacturers and chip foundaries often requires non-disclosure by the manufacturers – thus conflicting with the release of OSS driver code.
The blocking issues related to using Linux to do specialised professional audio are often not technical but much more to do with the Linux community’s doctrinaire, uncompromising attitude that the manufactures must swallow the virus form of GPL or loose the Linux market (Agnula for example). Often the manufacturers look at Linux’s 1% market share and unsurprisingly say Syonara.
I might be slightly but not much off with those latency times, and after some searching it seems that the low latency patch has been merged in 2.5 as part of the pre-empt patch (as lock-break???)… I think. I’m not completely sure. Someone correct me and update me, please. The subject fascinates me.
I run Gentoo with both pre-empt and low latency kernel patches, both are included in the very latest gentoo-sources (2.4.20-r2?), so it was not too hard to get working.
It’s the best distro for audio, clean and lean.
The pre-empt and low latency are seperate patches in 2.4, but I’m not really clear on what the difference between them is. I seem to remember hearing the pre-empt was the low latency patch done ‘correctly’, but is not as brutal as the low latency one. Having both at once works fine though.
It’s only the interface for jMAX that is java, the synth part is MAX. It has the advantage that you can run the interface on one machine and the processing on a few others.
You really have to subscribe to the forum to get the best out of it though, writing your own patches is fun up to a point, but for more complex stuff or instant results you need the patch library.
Csound has many graphical interfaces too, you don’t have to run it in text mode.
As far as what’s around now, works, looks ok, and more useful and accessable to most people… Here’s a couple that I like..
Freqtweak – Kinda like the NI Spectral Delay, but does pitch shifting too!
SpiralSynth Modular 2.1 – Yes, another modular, but easy to use and has analog like sequencers and a cool looper/scratcher.
I don’t think anyone is going to move their entire studio to Linux in the near future (unless some apps get ported or Ardour gets stable), but if you have the luxury of dedicating a box and audio hardware to it, there is much fun to be had.
I happen to think “Derive I” is quite funky. I’ve never been asked to record anything like it though.
Also, the m-audio range of cards all work fine under Linux, and any of the many other ICE1712 based cards.
If you have some decent D/As, almost anything with a digital out will do.
The reason there are not closed source drivers is more to do with the nascent state of ALSA than an innate disdain by the developers. It has not reached 1.0 yet, and the API still has the occasional revision. With an open driver, this just takes a tweak and a re-compile, with a closed source one, you rely on the manufacturer to keep up to speed.
Closed source firmware code for soundcards is used in ALSA without problems, as it does not effect the API. The reasons of the dislike of closed drivers is more technical than political.
I’m no audio pro but I know JMax is written in C and Java (C for daemons and Java for interfaces).
“ReZound … ReZound aims to be a stable, open source, and graphical audio file editor primarily for but not limited to the Linux operating system.”
And yes it works.
The problem with jmax is that it is not very compatible with its big brother, max/msp ( by the way, the synthesis part of jma is based on mas, not max. Max is the part to control RT stuff, Msp the DSP stuff ).
pD, Csound are NOT arguments to go to Linux, as it exists on windows or mac ( and by the way, pd is only about 0.41 version, so… )
About the drivers : of course, integrated alsa in 2.6 kernel will not make more drivers, but : all distro will install it by default, and if it is the standart, it will encourage everybody do go to Alsa, etc… RME has fairly good drivers on Linux, and my midiman delta44 ( semi pro card ) works perfectly with alsa ( I just had to compile it by myself, as alsaconf doesn’t work properly for me ). If midiman, RME, and later MOTU and presonus cards are supported, the problem of drivers will vanish.
The exemple of IRCAM is very biased, as it doesn’t reflect the majority of pro consumers ( by the way, there are far more apple boxes than Linux boxes for pro audio @ Ircam : I studied here during the last 6 months; there are still a few Next boxes in there ).
The installation IS a big problem : speak with the pro and semi pro. The Big problem is installation and configuration of softwares. A musician doesn’t want and doesn’t have to use line command, or a compiler.
I think you should try playing with xandros, it’s really worth it. They now have a much cheaper version that doesnt come with crossover and the like. If you’re like me and never use crossover anyways, it’s definately the best option… well, in my opinion anyways. Xandros isnt perfect, but is very simple and very comfortable, without sacrificing usability for the more technical userbase.
If I had money I’d get a mac. :/
I’m glad to see an article on Linux audio and it really does point out to people that Linux is not yet ready for it. The quality of apps and hardware support is lagging way behind that of both Microsoft and Apple.
The main problem with the Linux community in general is that most projects are started because the author needs that software for some reason, which means that you need a person who is interested in music and is a good enough programmer to start writing the software – rare as hen’s teeth if you ask me. This type of development has produced some great achievments, but as projects get bigger, people seem to cry bloat and fork the project or development slows down. You only have to look at Stalemeat to see how many projects have stopped being developed.
Also, for the system to be adopted in studios the whole OS/desktop etc needs to be totally idiot proof as most studios I’ve had experience in, the technicians are generally not PC experts – especially not Linux experts. I think Linux would be great hidden away inside a dedicated rack-mounted unit, I remember a similar thing some time ago that ran BeOS.
“Steve Cook, a DesktopLinux.com reader, relates his experience setting up Linux in the professional recording studio — including a switch to the Ogg Vorbis audio format. Increased productivity, greater stability, tighter security, lower costs, and a higher quality finished product are all reasons this station switched without missing a beat . . .”
I guess that most of us who have tried to record music on a Windoze system, have had all kinds of problems through the years. Problems with drivers, BSOD, lag, pops, MIDI/Audio getting out of sync, etc.
A Linux system/studio, that just works would be a dream…
A lot of the comments to this article remind me of what people used to say about the state of animation for Linux a few years ago. And now, it’s sudddenly taking over the high end… (I’m talking about the Dreamworks kind of thing)
A Linux system/studio, that just works would be a dream
Yes, isn’t it.
Runs sox on MDK 9.0 to convert an AU file to OGG.
>Channels: 1 Rate: 22050