At the Sounding Edge: Sounding Better All The Time

The second annual conference of Linux audio developers will be held in
Karlsruhe from April 28 through May 2. If you're in Germany during that
period and happen to be near Karlsruhe, you might want to stop by the
very cool
Zentrum fur Kunst und Medientechnologie
(ZKM) and check out some of the many lectures and presentations
about the cutting edge work going on in Linux audio software development.

For those of you who can't make it to the conference, I present a brief
State Of The Art message here in this month's installment of my At the
Sounding Edge series. There's so much going on in the Linux sound software
world that I can't hope to cover everything--not even all of the most
important developments--but hopefully I can give interested readers a
sense of what's happening and what directions we're taking.
ALSA and JACK
Without a doubt, the two most important development tracks are the
ALSA and
JACK projects. Both
supply the foundation for much of the most interesting application
development happening today, and support for ALSA/JACK rapidly is becoming
de rigeur in new sound software for Linux. I can't explore the technical
depths of these fundamental systems, but I'll briefly describe what
they are and why they are so important to the Linux audio development
and user communities.

I've covered ALSA in a previous column, so I'm going to be brief about it in this
article. ALSA is becoming a more familiar acronym, thanks to its inclusion
in the Linux kernel sources. The ALSA sound system effectively replaces
the older OSS/Free system, greatly expanding the sound services available
to developers and normal users. ALSA provides sound card drivers, a
library (libasound) for easy access to audio system functions and a
suite of handy utilities. Soundcard support ranges from consumer
grade cards to high-end professional digital audio boards. This advanced
hardware support is necessary for users wanting to use Linux in more
demanding audio production environments. Thanks to the efforts
of the ALSA team, we now can claim support for a small but significant
number of pro-audio boards. Without that support the Linux kernel sound
system would remain bound to consumer-grade audio, which may be effective
enough for casual use but certainly is insufficient for deployment in
a professional studio.

ALSA's utilities include tools for your soundcard mixer (alsamixer,
amixer), MIDI connectivity (aconnect), simple record and play programs
(aplay, arecord) and a variety of chipset-specific amenities. Almost
all are command-line tools, and although they may seem humble
mini-applications, they are quite powerful. I refer the interested
reader to the information available from the ALSA Web site; if you
already have ALSA installed, you can read the informative manual pages
(e.g., man aplay).

The list of the wonders of ALSA goes on, but this is neither the time or
place to explore them. Again, I refer the reader to the ALSA Web site
for the most up-to-date information regarding ALSA's current capabilities.

JACK is perhaps one of the most amazing pieces of software yet written
for the Linux audio system. The brainchild of Paul Davis (author of the
Ardour hard-disk recording system), JACK's capabilities include these
outstanding features :

• An easy method of adding audio support to new
  applications
• A robust professional-grade audio server designed especially
  for low-latency systems
• A transport control mechanism for synchronizing the
  activity of JACK-aware clients

The first feature is, of course, of special interest to programmers. Writing
directly for the ALSA API certainly is possible, but by comparison JACK is
a much simpler way to provide audio services to a client application. It
is no exaggeration to say that JACK has the potential to revolutionize
Linux audio development. The list of applications already supporting
it is an impressive testimony to its appeal and possibilities (see the
JACK Web site for a list of current JACK-aware software).

JACK is fairly agnostic about the underlying audio system, supporting
ALSA, OSS and PortAudio. It also is expanding to become a
multi-platform solution, with some deployment already possible under
the Mac OS X. Again, JACK relieves the developer of concerns about the
lower-level audio system, providing a simple and clean interface to the
underlying system without sacrificing the system's capabilities.

Inter-application communication is an important feature of both
JACK and ALSA. The ALSA sequencer API provides a way for compliant
MIDI applications to share I/O ports, permitting multiple access and
destinations. JACK does the same for audio, a rather more spectacular feat
of programming: MIDI streams are relatively thin compared to digital
audio and somewhat easier to manage and synchronize. Digital
audio is another kind of beast entirely, with inherent issues of
latency effecting the perfect synchronization of I/O streams. JACK has
been designed to address and resolve those issues, making it possible
to route and connect audio streams in much the same manner as ALSA's
sequencer handles MIDI data.

Another exciting aspect of JACK is its transport control. One of
the Holy Grails of audio development on any platform is the smooth
control (start/stop/continue) of various applications by using a master
controller. JACK can do this now with a number of applications, although not
all are completely transport-aware. An application must be written
with specific support for the JACK transport API, and only recently
has that interface stabilized to the point of wider acceptance and
deployment. Many well-known Linux sound and music applications already
support the JACK transport controls, and more are on the way.

Before leaving this section, I must note that the
OSS/Linux
driver package from 4Front Technologies still is alive and in active development. However, my focus here is on
free and open-source projects. I will cover OSS/Linux and other
commercial Linux audio ventures in a future column.
The Apps Scene
It might be hyperbole to say that the Linux audio software scene is
booming, but I feel safe saying that a sustained explosion
of activity in applications development has occurred over the past year. Some major
applications are evolving towards 1.0 releases, and others recently
reached their first birthdays. In addition, some new applications and ports have
appeared that show great promise along with their current usefulness.

Meanwhile, many Linux sound and music software stalwarts continue
their steady tread towards audio world domination. Hard-disk
recording continues to be almost the sole property of the
Ardour and
ecasound projects. Although Ardour draws
much attention and well-deserved praise, ecasound continues to
improve and remains the only command-line recording system capable
of professional use. Both applications are excellent examples of the
recent advances made by Linux sound software developers. Linux MIDI
sequencing software is well-represented by the
MusE,
rosegardenmusic.com
and
seq24 projects.

Other outstanding MIDI software includes the
JSynthLib
universal synthesizer patch editor/librarian and
the wonderful
KeyKit experimental composition
environment. Software synthesis languages are well-represented
by the venerable
Csound (now under the GPL/LGPL),
RTcmix and
Common Lisp
Music. Relatedly, development of the
Linux version of
SuperCollider3 is moving along briskly, with some GUI components already
available for alpha-testing.

Softsynths (standalone software synthesizers) abound under Linux, with outstanding
examples in
ZynAddSubFX,
ALSA Modular
Synth and
Qsynth, a GUI for the
fluidsynth
soundfont-based synthesizer. Soundfile editors
continue to be dominated by the
Audacity,
Snd,
ReZound
and
Sweep
projects. The
Hydrogen
drum pattern/song composer is clearly the
most actively developed application in its genre, and the
SoundTracker and
CheeseTracker
programs perpetuate the ancient art of music module tracking.

The
LADSPA plugin arsenal has expanded recently,
thanks to contributions from Tom Szilagyi, Tim Goetze and the
indefatigable Steve Harris.
VST/VSTi plugin support
also is expanding, with various loading mechanisms currently under
development. The Windows/Mac music software world increasingly is evolving
towards sophisticated plugin-driven frameworks, so support for VST/VSTi
plugins is of particular importance to musicians who would like to
switch platforms but don't want to lose their investment in commercial
audio software. However, it should be noted that many LADSPA plugins
are as powerful and sophisticated as their VST counterparts. Having
both LADSPA and VST/VSTi plugins running under Linux is simply
a win for everyone.

Lest casual desktop audiophiles feel left out of this assessment,
excellent A/V file and stream players can be found in
XMMS,
AlsaPlayer,
xine and
MPlayer. Those last two
players also include versions of themselves as browser plugins with impressive support for the array
of video formats found on the Web (yes, you can now watch Quicktime
trailers on-line in Mozilla). In addition, Linux players and plugins for the
Real and Flash formats are available from their parent companies.

It also is worthwhile to consider briefly where Linux sound and music
software is weak, especially when compared with what's available to the
Windows/Mac musician. Of course, by design Linux audio developers have
steered clear of attempting monolithic applications such as Cubase or
Logic Audio. However, many of the functions of those applications can
be replaced by running a JACK-connected LADSPA-savvy set of software
such as Ardour, Hydrogen, Rosegarden and ZynAddSubFX. More significant
holes exist in the domains of music notation software, loop composition
software (such as Acid) and advanced software samplers. Again, depending
upon the complexity required, it may be possible to combine some Linux
software to achieve some of the Win/Mac software's functions, but users
coming from the Win/Mac world may be put off by the complication of
the combined applications approach.

Regarding music notation software,
Linux has nothing like Finale, although the combination of
NoteEdit with
LilyPond has great potential; LilyPond produces superb
print output. Regarding Acid-like programs, it has been pointed out
that Ardour can be used in similar fashion to some extent, but those
capabilities would be much enhanced by the incorporation of
more sophisticated controls over looping sound (e.g., pitch-shifting,
time-stretching, real-time loop-point edits). Nevertheless, no Linux
application currently compares to Acid, and the same must be admitted
for sampling applications, such as Gigasampler. Projects are under way to
design such software for Linux, but nothing yet exists to match them.

I am planning another article to discuss Windows and Mac music software
that runs under emulation in Linux, so I must excuse myself for not
considering here audio applications that can be run under WINE, VMware,
ARDI Executor or other Mac and Windows emulation environments.
Cards & Drivers
Linux still suffers from a lack of support by some of the better-known
hardware manufacturers. Some companies have indicated that Linux simply
isn't in their development radar. Until the numbers urge them, they
have no interest in either developing drivers themselves or giving
the necessary specifications to third-party developers, such as the ALSA
team. As a result, Linux pro-audio hardware support currently depends
almost exclusively on the boards and cards from RME and
M-audio.

The news is better for desktop audio enthusiasts, with many popular
soundcard chipsets supported by ALSA drivers. Support for the Creative
Labs SBLive is especially good, with working 5.1 sound, hardware synth
support and digital audio output. Laptop users may be happy to note
that USB sound and MIDI device support has improved, and useful PCMCIA
audio cards can be purchased from
CoreSound, RME
and
DigiGram.
The Short Summary
Just as Linux itself no longer is regarded as merely an interesting
hack, Linux audio software also is growing into what I call its mature
adolescence. Even the often-dreaded installation and configuration details
have been largely resolved by the
AGNULA distributions
and the
Planet
CCRMA packages. Plus, live CDs, such as
dyne:bolic let anyone
check out the multimedia possibilities of Linux without installing
anything.

Alas, I've touched on only a few interesting developments going on in the
Linux audio world. I'm excited to learn more about the most recent Linux
audio evolutions in Karlsruhe, so be sure to check out next month's column
for an update on the latest activity in Linux sound and music software.

Dave Phillips (dlphilp@bright.net) is a musician, teacher and
writer living in Findlay, Ohio. He has been an active member of the
Linux audio community since his first contact with Linux in 1995. He is
the author of The Book of Linux Music & Sound, as well as numerous
articles in Linux Journal.