|
| News | Profile | Code | Photography | Looking Glass | Projects | System Statistics | Uncategorized |
Blog |
PM Vibes has some great trance mixes available for download. Unfortunately, they're provided in the form of one large MP3 file with a cue sheet describing the track layout for burning to a recordable CD:
-rw-r--r-- 1 prox prox 543 2009-07-01 11:22 PM - Dream Melodies volume 10 CD.cue -rw-r--r-- 1 prox prox 424471 2009-06-30 22:15 PM - Dream Melodies volume 10 cover.jpg -rw-r--r-- 1 prox prox 115071104 2009-06-17 12:04 PM - Dream Melodies volume 10.mp3 -rw-r--r-- 1 prox prox 2108 2009-07-04 16:23 PM - Dream Melodies volume 10 SONGS.cue -rw-r--r-- 1 prox prox 1184 2009-06-17 15:58 PM - Dream Melodies volume 10 tracklist.txt
To make matters worse, there seems to be absolutely no way of burning or converting this on Debian GNU/Linux without some [re]compilation.
The packages that should provide this functionality include the following:
Every one of the above applications (ok, so Gnomebaker is just a cdrdao GUI front-end) claim to read cue files, but the Debian packages don't include MP3 support. You'll get something like this:
% cdrdao write --overburn -v 2 --device /dev/scd0 --speed 16 "PM - Dream Melodies volume 10 SONGS.cue" Cdrdao version 1.2.2 - (C) Andreas Mueller <andreas@daneb.de> SCSI interface library - (C) Joerg Schilling Paranoia DAE library - (C) Monty Check http://cdrdao.sourceforge.net/drives.html#dt for current driver tables. ERROR: Can't read file "PM - Dream Melodies volume 10.mp3": cdrdao was compiled without MP3 support. ERROR: PM - Dream Melodies volume 10 SONGS.cue:165: Cannot determine length of track data specification.
So it would seem the only way to burn such a beast is to compile cdrdao (or wodim, whatever) by hand. I needed to grab libmad0-dev, and configure cdrdao via the following:
% ./configure --with-mp3-support
This resulted in the following summary, which ended up building a cdrdao binary that includes MP3 support:
------------------------------------------------------
Building scsilib : no (using native interface)
Building pccts : yes
Building cdrdao : yes
OGG support : yes
MP3 support : yes
Building toc2cue : yes
Building cue2toc : yes
Building toc2mp3 : yes
Building gcdmaster : yes
------------------------------------------------------
However, the fun wasn't over, yet! The cue file still caused problems:
% $HOME/src/cdrdao-1.2.3/dao/cdrdao write -v 2 --device /dev/scd0 --speed 16 *SONGS*cue Cdrdao version 1.2.3 - (C) Andreas Mueller <andreas@daneb.de> ERROR: Can't read file "PM - Dream Melodies volume 10.mp3": cdrdao was compiled without MP3 support. ERROR: PM - Dream Melodies volume 10 SONGS.cue:165: Cannot determine length of track data specification.
You'll notice that I got the same error earlier, too. Apparently there's a bug in cdrdao that was introduced by a patch (that subsequently made its way into Debian, too - more strikes against that package!) related to reading of track lengths in TOC files, reversing changes made by upstream. Uh, yeah, this breaks stuff. An Ubuntu bug #490292 was filed against the cdrdao package about this.
I commented out the block of code in TocParser.g, and recompiled. Finally, success.
Things are really not as simple as they should be, sometimes.
I'd submit bug reports about this, but I'm fuzzy on the complexity of MP3 licensing - the Debian developers may have no choice but to remove MP3 support from these packages. I did post a message to the debian-user mailing list about this.
I'm starting to get annoyed at electronics sporting power buttons that require the user to hold them down for several seconds before anything happens. Computers, wireless headphones, Nokia cellphones, the list goes on. Most of them are recessed buttons, too, so the chance of something accidentally "bumping" the button is slim at best. When I want something turned off, I want it to comply immediately!
It also annoys me off that these buttons can all be disabled in software (which also makes them susceptible to software bugs).
Am I the only one perturbed by this?
Yes, I know it's not called Core i9 anymore, but I will continue to refer to the mammoth upgrade of my workstation as a Core i9 Upgrade.
A couple folks have been asking about this, and since I couldn't figure out how to make my Newegg public wish list viweable by people who don't have a Newegg account, I've reproduced it here with comments:
Intel Core i7-980X LGA1366 CPU (Retail Box)
Note: This isn't out yet. It's supposed to be out by March 16th, 2010. Yes, six cores with HT.
Intel BOXDX58SO LGA 1366 Intel X58 ATX Intel Motherboard - Retail
Model #:BOXDX58SO
Item #:N82E16813121361
GIGABYTE GV-N98TSL-1GI GeForce 9800 GT 1GB Silent Cell 256-bit GDDR3 PCI Express 2.0 x16 HDCP Ready SLI Supported Video Card - Retail
Model #:GV-N98TSL-1GI
Item #:N82E16814125282
Note: I'm only getting this card because it lacks a fan. I have terrible luck with fans, they all die on me after a year or two (yes, I change the air filter in my HVAC unit frequently) for some reason, and I can't stand the noise they generate.
DViCO FusionHDTV7 Dual Express DUAL HDTV/Analog TV Tuner Card - Retail
Model #:FusionHDTV7 Dual Exp
Item #:N82E16815135001
Note: I'm not buying this from Newegg. I might not buy this at all - I already have the FusionHDTV5 RT GOLD card (I use the 8VSB tuner with MythTV for crystal-clear OTA broadcasts), which is a single-tuner classic PCI version of this card, and since the Intel motherboard has only one PCI slot, I have to decide if I'm going to stick with the FusionHDTV5 or my old SB Live! card. If I ditch the SB Live! card (that has served me well for 11+ years, and does Dolby Pro Logic II decoding in hardware, or something) then I'll stick with the FusionHDTV5.
OCZ Colossus Series OCZSSD2-1CLS250G 3.5" 250GB SATA II MLC Internal Solid State Drive (SSD) - Retail
Model #:OCZSSD2-1CLS250G
Item #:N82E16820227473
G.SKILL Ripjaws Series 4GB 240-Pin DDR3 SDRAM DDR3 1600 (PC3 12800) Desktop Memory Model F3-12800CL9S-4GBRL - Retail
Model #:F3-12800CL9S-4GBRL
Item #:N82E16820231313
Quantity: 4x (yep, 16GiB!)
Note: There aren't many 4GiB DDR3 RAM chips around.. I'd be completely satisfied to get ones that don't have red jaw looking pieces of plastic attached to the PCBs, but I don't have much of a choice at this point.
I'm using my existing case, power supply, and Blu-ray/DVD burner optical drive. I guess I have to ditch the 3.5" floppy drive, since there's no floppy controller on the DX58SO..
I've been using Asterisk with BroadVoice for a little over six months, now. Other than occasional network latency, I've been happy with the setup.
However, sometime in the last few weeks, incoming calls stopped working. I don't rely on inbound calling too much, so it took me awhile to notice it was broken. Outgoing calls worked just fine, but inbound calls went straight to BV voicemail. Here's what the SIP traffic looked like:
< INVITE sip:xxxxxxxxxx@10.3.5.1:5060 SIP/2.0 < Call-ID: 97037a-97@147.135.0.128 < CSeq: 1 INVITE < From: "Unavailable"<sip:147.135.0.128>;tag=479b < To: "Mark Kamichoff"<sip:s@10.3.5.1> < Via: SIP/2.0/UDP 147.135.0.128:5060;branch=z9hG4bK-BroadWorks.192.168.0.3-192.168.0.128V5060-0-470344215-1909890038-1265808557101- < Contact: <sip:147.135.0.128:5060> < Supported: 100rel < Max-Forwards: 69 < Content-Length: 309 < Content-Type: application/sdp < < v=0 < o=2475098871 10 10 IN IP4 147.135.2.247 < s=- < c=IN IP4 147.135.2.248 < t=0 0 < m=audio 21338 RTP/AVP 0 8 18 96 97 101 < a=rtpmap:0 PCMU/8000 < a=rtpmap:8 PCMA/8000 < a=rtpmap:18 G729/8000 < a=fmtp:18 annexb=no < a=rtpmap:96 iLBC/8000 < a=fmtp:96 mode=30 < a=rtpmap:97 t38/8000 < a=rtpmap:101 telephone-event/8000 > SIP/2.0 401 Unauthorized > Via: SIP/2.0/UDP 147.135.0.128:5060;branch=z9hG4bK-BroadWorks.192.168.0.3-192.168.0.128V5060-0-470344215-1909890038-1265808557101-;received=147.135.0.128 > From: "Unavailable"<sip:147.135.0.128>;tag=479b > To: "Mark Kamichoff"<sip:s@10.3.5.1>;tag=as4c6fddbc > Call-ID: 97037a-97@147.135.0.128 > CSeq: 1 INVITE > Server: Asterisk PBX 1.6.2.0~rc7-1 > Allow: INVITE, ACK, CANCEL, OPTIONS, BYE, REFER, SUBSCRIBE, NOTIFY, INFO > Supported: replaces, timer > WWW-Authenticate: Digest algorithm=MD5, realm="asterisk", nonce="758e4a53" > Content-Length: 0 < ACK sip:xxxxxxxxxx@10.3.5.1:5060 SIP/2.0 < Call-ID: 97037a-97@147.135.0.128 < CSeq: 1 ACK < From: "Unavailable"<sip:147.135.0.128>;tag=479b < To: "Mark Kamichoff"<sip:s@10.3.5.1>;tag=as4c6fddbc < Via: SIP/2.0/UDP 147.135.0.128:5060;branch=z9hG4bK-BroadWorks.192.168.0.3-192.168.0.128V5060-0-470344215-1909890038-1265808557101- < Max-Forwards: 70 < Content-Length: 0
BV was sending Asterisk an invite, Asterisk was challenging it (expecting a secret), but BV was giving up.
I found this thread back from 2008, which described a similar problem that ultimately was blamed on BV. However, I'm using Asterisk 1.6 (1.6.2.0-1), not 1.4, so I figured it wasn't really relevant.
Upon closer inspection, it appeared that the insecure option I was using was deprecated:
[Feb 11 11:39:22] WARNING[18106]: chan_sip.c:23160 set_insecure_flags: Unknown insecure mode 'very' on line 1139
I changed that to insecure=port, but that didn't fix the problem, either. Apparently insecure=very is the equivalent of insecure=port,invite, meaning Asterisk won't try to authenticate incoming calls (invites).
This ended up working, and I have the following as my SIP peer for BV, now:
[sip.broadvoice.com] type=peer user=phone host=sip.broadvoice.com fromdomain=sip.broadvoice.com fromuser=xxxxxxxxxx secret=yyyyyyyyyy username=xxxxxxxxxx insecure=port,invite context=external authname=xxxxxxxxxx dtmfmode=inband dtmf=inband canreinvite=no
I guess it was due to an upgrade I did awhile back. Note to self: check Asterisk after doing dist-upgrades.
Apparently the math question for comments (to prove you are not a stupid spam bot) on my blog hasn't been working consistently:
14:12 < cnj> your math is broken, prox. 14:27 < cnj> what would you do to imporve last.fm... 14:33 < prox> Hm? 14:33 < prox> What math? 14:35 < cnj> your news 14:35 < cnj> It also doesn't transfer your entry when you click retry. 14:35 < cnj> but it's ok, cause it won't actually post.
Apparently I had some floating-point precision issues that started causing problems when the user-submitted math answer was compared with the generated one. I was seeing things like 11.1 != 11.1. I hate floating-point data types. So, I just dropped the decimal point and now the math questions are generated by two calls to rand(1,100). It seems to work.
Also, I fixed the form so it keeps existing input if you enter the math question wrong, and fixed a few miscellaneous bugs. I probably should have done this sooner.
In other news, I had my 2009 review, today. Apparently I've been doing something right, since my boss had some good things to say about my performance over the last year. Now, we'll see if I can manage to not slack off in 2010. Fingers crossed!
I was just at Subway, ordering some lunch. The checkout went something like the following:
Subway Employee: Combo, sir?
Me: I'll just have the sandwich, an orange juice, one peanut butter cookie, and one chocolate chip cookie.
Subway Employee: <thinks>
Subway Employee: Not a combo!
For a second, there, I felt like I was playing a video game, and doing a terrible job at it.
After that dialog, she asked me what kind of sandwich she had just put in the bag, for me, like they do every time. Why don't they communicate in some way with the sandwich artist (that is apparently the correct title) so that they don't have to ask the customer this, every time? If I wanted to save some money, I probably could have said I got the cheapest sandwich! Really, a sticker or something would be easy. I suppose the sticker or other communication method probably costs more than the loss incurred by customers lying and saying they got a cheaper sandwich. Maybe not? Somebody should do a study.
Anyway, jalapeno peppers + lots of black pepper == hot sandwich.
Gah, the Mecklenburg County Aquatic Center schedule is annoying me, lately:
*5 - TCS Swim Meet - Entire Facility Closed - 3pm - 9:30pm
*6 - 7 - TCS Swim Meet - Entire Facility Closed
It's a good thing I kept around my YMCA membership (even though their pool is closed all too frequently). It's ashame that there isn't a non-crowded pool in the south Charlotte area that is consistently open. Whatever.
I hate it.
Here's my dilemma:
A few weeks ago my parents sent me their old camcorder with a dozen tapes of ancient home videos. After playing with my video capture card (DViCO FusionHDTV5 RT Gold - using the composite input) and MEncoder, I finally got a step-by-step process down for each of the tapes. It went like this:
Press play on camcorder and then immediately execute the following command:
sudo nice -n -10 mencoder -ovc lavc -lavcopts vcodec=ffv1 -oac mp3lame -lameopts cbr:br=224:mode=3 -o raw.avi -tv input=1:normid=0 tv://1
The above encodes to ffv1 (lossless codec) the video coming in from the composite input of my capture card. The nice command tells it to run at a bit higher priority than normal - mostly because I want to keep using my workstation when this is recording, but not have any user processes (yeah, Flash video) take away from the encoding process.
Next, I do the following to do a two-pass encode of the video with H.264 while keeping the audio intact (we converted to MP3 in the previous step):
mencoder -ovc x264 -vf pp=lb/ha/va/dr,hqdn3d -x264encopts pass=1:bitrate=3000 -oac copy -o /dev/null raw.avi
mencoder -ovc x264 -vf pp=lb/ha/va/dr,hqdn3d -x264encopts pass=2:bitrate=3000 -oac copy -o 1996.avi raw.avi
The pp=lb/ha/va/dr,hqdn3d does some magic by deinterlacing the video and making it look nice. The output is an AVI container with an H.264 video track and an MP3 audio track.
So far, so good, right? I did that for all twelve videos, and now am running into a roadblock.
I want to convert all of those to MP4, so QuickTime (yes, my parents have an iMac) can play it. MEncoder doesn't support MP4 directly, so I have to do an offline conversion. Normally, I'd just do the following with FFmpeg:
ffmpeg -i 1996.avi -f mp4 -vcodec copy -acodec copy 1996.mp4
MPlayer will play this fine (well, not fine - when seeking, MPlayer can't find keyframes, so starts with a whole bunch of crap on the screen, but this gets better once a keyframe is encountered), but.. QuickTime hates it. The video stutters horribly, for some reason.
So, I try Avidemux, a video editor I use for a number of other things and normally works very well. Unfortunately, the MP4 it created doesn't even open in QuickTime! Turns out this is a known issue. Avidemux is apparently creating a textbook MP4, but somehow QuickTime didn't read the textbook. Fail.
I search around a little bit more, and stumble across MP4Box (part of the gpac package in Debian provided by the Debian Multimedia repository). It required me to split the AVI into its tracks, and then use those to encode, while specifying the framerate (which made sense, since the FPS is in the container):
MP4Box -aviraw video 1996.avi
MP4Box -aviraw audio 1996.avi
MP4Box -fps 29.970 -add 1996_audio.mp3 -add 1996_video.h264 1996.mp4
This doesn't work right, for some reason. MP4Box doesn't listen to the FPS I'm giving it, and defaults to 25! Audio and video become desynchronized and it all goes downhill. I tried -add 1996_video.h264:fps=29.97, too, not really understanding the command-line arguments, but that didn't do anything, either. I suspect a bug, but I'm not going to bother with that.
I searched around a little more, and I found a Windows utility called MP4Creator that looked like it might be able to do the trick. I got mixed results, mostly because each time I ran it, I got a different file size! I also didn't know how to add two tracks, but apparently if you run it twice, it re-interlaves the second track into the existing MP4 (strange). Here's what I did:
mp4creator -rate=29.97 -c 1996_video.h264 1996.mp4
mp4creator -c 1996_audio.mp3 1996.mp4
Well, this plays perfectly with MPlayer (no problems with seeking!) but, now, we're back to the file not even opening with QuickTime.
What is going on, here? Is there something else I should be using? Re-encoding is not an option (and it shouldn't need to be - I have the raw H.264 and MP3 files, and I know the framerate) at this point.
Displaying page 18 of 121 of 965 results
     |
This HTML for this page was generated in 0.007 seconds. |