Art
Computing
Contact info
CV
Electronics
Geek culture
Home brewing
Math
Music
Physics
Rants
Ψ/Φ
Style
Teaching
Theatre
WHAT'S UP
<23:34 EEST>
Some days ago I finally went Ext4 with the largest partition of gaga. I converted it from Ext3 as instructed by
Kernelnewbies. Things have
worked smoothly so far, and since the Ext* family is all about backwards
compatibility and stability, I'll probably use it as the root FS of my next
Linux install, whatever that will be.
In fact, I've had some trouble with my laptop's JFS root. After the machine crashes for whatever reason -- for example, a hard reboot when NFS shares fail to umount -- it is often unable to boot properly due to a condition I call "mtab hell". The root filesystem is mounted readonly, and it cannot be remounted as RW, because /etc/mtab is unwritable for obvious reasons. I'm sure this is a common situation and the init scripts generally blank the mtab at some point. As this is the only machine where I've encountered the situation, I suspect the reason is really with JFS. Another pointer towards the JFS problem is that when I fix the situation with a live CD, the root partition cannot be mounted unless explicitly fscked first.
It's really unfortunate, as I've regarded JFS the most stable filesystem until now, and it's even among the fastest. It peers with Reiser3, and while XFS might be faster on good hardware, I've found it remarkably unreliable. The last time I used XFS on Linux it had the well-known bug of not knowing when the disk was full. It would produce some really funky errors when it had tried to write too much data.
Some people might wonder why I bother with this geek stuff in my spare time, even while I have a rather technical day job. One explanation is that I'm really bad at mode switching. I'd really like to be writing new songs, for example, but there's too little spare time and space dedicate for something creative. So instead, I try to enjoy being a full-time Linux and hardware geek. I guess this also gives me some leeway, for example checking out important technology news during working hours.
<20:55 EEST>
In most discussions on multichannel audio, I've been a rather fierce
promoter of basic 2-channel systems. (Note that it's not 'stereo' because
that means 'spatial'.) I've used the simple information-theoretical argument
that everything you listen to eventually gets downmixed to two input
channels into your brain, and no processing between a varying number of
intermediate channels can improve the sound.
However, I've also been aware of the problems of 2-channel all this time. Getting those two recorded channels intact and distinct into their respective ears is by no means easy. One key problem is with those sounds that are panned to the middle, such as most vocals and bass. As they have equal volumes in both channels, you have to listen to them at the exact centerline to get the desired balance. Off to a side a little, and you lose the illusion of the sound center. This is understandable purely from the volume standpoint, but there are further and nastier effects from the different phases. Some frequencies are cancelled, others reinforced.
Thus in my current 3.0 setup, I've found that the center speaker makes a world of difference. Most of the sound is actually coming from the center, making left and right effects all the more special. For music I've used the Dolby Pro Logic II setting that, while doing some nasty processing to the original 2-channel recording, makes great use of the center speaker. Actually, the music mode keeps L and R intact, but of course the overall soundscape is altered.
Without knowing the exact details, I wonder how DPL II deals with the problem of interference -- the above issue with phases when the same frequency is played from different sources. There should probably be some filtering to make the channels distinct even in frequency space, enhancing the illusion that different instruments are played from different speakers. On the other hand, many musical spectra are surprisingly wide, and splitting that of a single instrument over multiple speakers might sound weird, perhaps to the point of a novel effect. Of course that kind of splitting is already done between the elements of a speaker cabinet, but with the intention of keeping them as close as possible.
While this may sound (pun intended) trivial to some people, I really wanted to point out the importance of the center channel. A lot of the hype surrounding (p.i.) multichannel audio seems to be about the surround speakers, the magical effect of hearing things behind you. Also a lot of setups have large and sophisticated L and R speakers, with the rest of the system some smaller satellites. The empirical fact, however, seems to be that most of the sound we listen to is at the very center, with sounds from other directions providing additional clues about the source.
This is the key idea in Ortoperspekta, where there's only one front speaker and two surround channels. I've also experimented with it and I find the concept interesting; as I understand it, OP makes use of interference to recreate the R and L in your ears approximately, even if you're not sitting at the sweet spot. However, it is also not without problems. A true multichannel system, with separately recorded tracks, could in principle be even better. But with the usual 2-channel music you have to resort to some kind of processing to disperse it over a wider soundspace. I'm generally against all unnecessary processing, but since there's no correct way of doing it, there are no wrongs either, whether it's OP or 3.0. I've enjoyed both and I'll probably keep on experimenting further.
<23:24 EEST>
Today's sync was solving a major boot issue in two wholly different cases:
first a BIOS update in my work laptop made
displays work when booting docked, then gaga
finally got into the shiny new world of the 2.6.28 kernel series. I have no
idea what actually made the latter. Lately yaboot had been acting up, and I
had been a little careless with my cross-distcc
setup, but today something had gone right.
On another note, I've been meaning to write about some wonderful new audio experiences with my digital 3.0 setup, but I haven't found any creative juices left after workdays. Maybe some day I'll be free to do that.
<23:29 EEST>
After nearly four years with this laptop, I've found out something new about
the hardware, which happens to be a connector type hitherto unbeknownst to
me. I was generally aware of the digital audio output, but it didn't work as
I'd expected, plugging the electrical contact into a digital RCA input at
the amplifier. The digital output turned out to be optical, embedded in the
same 3.5 mm jack socket, and connecting to an ordinary Toslink fibre with a
simple adapter. Of course, this was completely undocumented, with only vague
mentions of S/PDIF without regard to the physical medium.
Compared to that realization, finding the cable and the adapter has been a breeze, as well as the software side. The MPlayer man page pointed to a shortcut "-ao alsa:device=spdif", and with further options you can direct the raw AC3 or DTS stream into the amplifier.
<23:44 EEST>
Today I got myself three more of the JPW Mini Monitors that I already have
two of, plus a brand new, basic surround amplifier, the Pioneer VSX-418. The
grand plan with my recent acquisitions is that I don't want any duplicates
if I ever move in with myself, but a total of five identical award-winning
speakers is certainly no excess. There's still room for a decent screen and
a subwoofer.
My 'inexpensive and practical dream' system would simply have more channels in a Gainclone, and none of this processing crap. Particularly when the six channels of film audio are already stored separately. It's the time and space for building that I don't have right now, which is why I opted for a really basic surround amp. With my attitude towards this kind of consumer electronics, anything more expensive would be a waste, and I'll always have my current GainHog for two just about perfect channels.