EtreeWiki - Sector Boundary Discussion

Recent discussion with some helpful links even beyond SBEs:
http://forums.etree.org/viewtopic.php?t=1393&highlight=

An earlier essay, with more emotional discussion (note message datestamps in links):

The importance of sector boundaries has been discussed in much detail at the following locations:

1) http://db.etree.org/shninfo_detail.php?shnid=11614
2) http://groups.msn.com/gdlive/general.msnw?action=get_message&ID_Message=21054&ShowDelete=0&CDir=-2

There are a few important comments made on these pages.

The first, by Jay Serafin (a contributor of numerous seeds via the GDLive server), outlines his feelings regarding the issue. To summarize, Jay does not feel that tracking shows on sector boundaries is important. Here is his full commentary:
"Look at the GDLive message board for the once and for all answers to the SBE "problem". The "Mixdown ADAT" was taking the pre-mixed (2-track) show, and giving me FOUR REDUNDANT COPIES in case of a problem on the ADAT tape. "100 digital realm" means that from the entry point into any of the three editing computers, to the both Harpoon-propriatary digital input sound card, to the digital mixer, to the Harpoon editing software, there are NO analog components involved in editing and processing this show. If you say you've e-mailed me, but I don't answer, it's because I didn't get the e-mail! I answer ANY AND ALL non-spam e-mail questions posed to me. So, don't lie about things. To save you the trouble of going and reading the GDLive post, here it is: "I am going to explain this for the LAST TIME! (And I've done so in MANY private e-mails, despite people saying I do not reply to them). Since these shows are sent to me for (usually) touch-up work, editing, noise reduction of all types (not just analog tape hiss), and other processes, my goal is to make the show sound as natural as possible, almost to the point of "being there". Most of the time, this is able to be accomplished. Sometimes, it is not. Having said that, I work on each track/song indivudually. This allows me much more freedon and saves much more time than working on it as an "entire set" or "entire show". When you to and see a band play, they start EACH SONG on the first note. This is NOT POSSIBLE if I were to cut the songs, as you folks say, "on the boundary". Depending where the boundary actually is, the song can start too soon, or too late. If it begins too late, than you get the first part (note, word, whatever) chopped off. This is not how my releases are meant to be. I work long and hard on these shows. I am more of an "archivist" in the work I do. Because of the way people burn their CD's, the software used, the burners used, and I could go on, some people get the "click" between tracks, some people do not. THIS IS NOT MY CONCERN! This may sound harsh and cruel, but people say "gimme, gimme, gimme", and expect everything to be handed to them on a silver platter. This is NOT how it works here! Each .shn file is a complete song/track onto itself. You can take ANY individual .shn file from ANY of the shows I've worked on, decompress it, and you will have the song/track play 0.04 to 0.06 seconds from the very first note of the song (this is what is also used in mastering houses world-wide... it is done because many CD players will not instantly produce output. They need a little "starting room", and this very small, and unable-to-be-heard "pre-start" (which is actually the very end of the last track/song) will give any CD player that chance to get up to speed, and to begin capturing the data from the audio CD people burn. So, if you only wanted to hear Me & My Uncle from one show I've released, if you download that .shn file, you're going to get it... not starting too soon or too late, but right on the money! If people don't like having to deal with the SBE problem, here's my solution: Take the show, uncompress the files, stitch the tracks together, take whatever your favorite program is, and let your choice of software break the show into the sections it wants to do. There it is... problem solved! I release the shows, so that people can go and obtain a song just by uncompressing ONE FILE, not TWO files! If the original audio files were "cut on boundary", you'd need the preceeding file (if the song you wanted started "late", due to the SBE program), or you'd need the file after the one you wanted (if the SBE cut the song "too early" at it's end). So now, you have MORE WORK TO DO! I am not forcing anyone to listen to shows made with this SBE program, that SBE program, etc. A commercial studio doesn't do this, and I don't either. The mastering houses have ways of taking care of this by means of million dollar hardware/software combinations when they make the pressing copy for future CD discs. You can take and re-do the shows I release in ANY FASHION YOU WANT! Want a single MP3 song? You can take the .shn file, and make a pure MP3 of a song just by downloading the one file! Again, you don't need TWO files! Finally, I find it very unusual that so many people have problem with my shows. My AUDIO MASTERS which I use for making dubs (audio files, not .shns) are made from the MASTER .SHN CD! I simply uncompress each file, use CDRWIN to burn the CD at 1x speed (which is the MOST RELIABLE WAY to get the proper burn depth, and without introducing jitter and worse, due to mild to severe vibrations when a disc spins at anything over 4x). The pits are burned to the correct depth, and can be read by high-speed CD-R drives, and ripped without problems. If one burns at anything over 4x, the pits may not be deep enough, and when the disc is ripped, diginoise can be introduced into the ripped track. If people are too lazy to do a little work, that's not my problem. I'm giving people shows which can be used to make compilations, etc., with the least amount of work. And that is because of each file contain ONE SONG PER FILE. No "undercutting", no "start of the next song at the end of the track" being in there... it is just like taking your home/portable CD player, and choosing Track 04 (for example)... you certainly expect to only hear Track 04, not the first part cut off, or part of Track 05! Use whatever program suits your needs to "fix the problem". But, the way I release my shows, there IS NO PROBLEM! You can take (just like I do) ANY show I've released, decompress the .shn files for (example) Disc 1 of 3, and then using good software (the 1x version of CDRWIN is FREE, by the way!), burn the disc. But people are looking for shortcuts, and ways to save their own precious time. Sorry, I don't work that way. The releases I put out there are made in a specific way, for a specific reason and purpose. If they're going to be vined, passed to others, whatever, that is NOT MY CONCERN! My concern is that the individual person who is personally downloading the show for their OWN enjoyment; or to have a show where (s)he can download and listen to a single song, or even multiple songs, using mkwACT, and their favorite player software (or burning their own personal CD-R). That's it. Cut and dried. It's funny how many people bitch and moan about this SBE problem, yet so many others write and tell me how good the show is! Hey, I could just as easily put the .wav files up here on GDLive, and let you folks run wild with it! But no, people wanted them compressed to save time. So, you have to live with the limitations of technology, as do I! Sorry if this post sounds bitchy, high-handed, egotistical, call it what you will. You wanted an answer, you got the answer! Maybe it's not the answer you wanted. It can take my anywhere from one week to one month to release a high-quality audio show, as I have to balance family responsibilities, my physical limitations of being unable to sit for more than 30-45 minutes, household chores, social responsibilities, and more. You folks can't take the hour or two to "make the show work like you want it to behave"? Sorry, but to me, people who don't want to do this appear greedy and/or lazy. Gee, what if don't use YOUR favorite program to cut the tracks? I guess I'm still "wrong", even if it's not right! Or, we can simply end the whole problem right here and now: I just won't release any more of the 309 shows left to be worked on. I'll throw the media into the trash I'm sitting here watching footage on every friggin' channel, watching my brother firefighters and paramedics die needlessly. Can't excape it today. But, popi got this going, and it deserves a reply. For those of you who say "Jay doesn't e-mail me back", I say you are a bold-faced liar! I answer EVERY piece of non-spam e-mail sent to me!!! People ask me questions, I give them answers. Always have done, always will do." "

The second, a statement coroborated by several individuals including Caleb Epstein and RMS, was made by Ghost. It follows:

"i think the problem here is that jay doesn't seem to really understand what Sector Boundary Errors are, and how they can be avoided. If Jay would take 2 seconds to run his files through shntool once he finished them they would have no errors (resulting in no clicks in between tracks), and there would be no problem. There is no way in hell that fixing a sector boundary error will have any discernible impact on where the song starts. It's just too small of a shift. For those of you who argue that we should just be happy he does this to begin with i say you are wrong. he is being stubborn about doing 2 seconds worth of work that would prevent us from having multiple md5s of a show in circulation. "

RMS made a similar statement with a link to more information about sector boundaries:

"I'll give Jay the benefit of the doubt here, but I don't understand how a "Chief Audio Engineer" can be so ignorant of a basic wave editing principle. You can check out http://www.mrichter.com/cdr/primer/blocks.htm for some more information about CD writing. Here's the short story: the redbook standard (which Jay clearly knows about and respects given his refusal to track shows for 80 minute discs) defines 75 blocks in each second of music. There are 588 samples per block... if you split tracks off of sector boundaries, the rest of the 588 sample block is padded with silence or other noise (depending on several variables). Thus, when burning live shows DAO there is either a break/pause between songs or a "click". To remedy this, you need to cut your tracks on sector boundaries. At most, this will mean moving the track markings 1/150th of a second. If that means you don't start your track at the "actual beginning of the song", I say "not true". And as the page I referenced points out, "Of course, that means your split may be moved. If you can hear the difference when it's shifted no more than 1/150th of a second, try analogue recording." But, as Jay has pointed out so vehemently, he's a 100% digital kind of guy. So Jay: cut your shows on sector boundaries! Why don't you? There is zero reason not to do the community this favor. Even if you really do just want one person to enjoy the show themselves, they'll get clicks or silence between tracks when they burn their "own CD-R copy". That will hamper their enjoyment of the show, which seems to be your goal. Again, I don't have any beef with you. I think it's great that you and your "Marin County Source" contribute to the community. At the same time, members of a community share some basic rules of etiquette. In our community, one of those "rules" is that seeds should be cut on sector boundaries. Simple enough, right? If you have any questions about ways to enable your code named "Harpoon" editing software to cut tracks on sector boundaries, feel free to post here and someone will help you out. "

Hopefully this will help clarify the importance of sector boundaries when seeding shows. Remember that your seed could be enjoyed by hundreds or even thousands of listeners who deserve only the best. Take your time when seeding and do it right - We *all* greatly appreciate it!

someone else added:
Here lies the problem with Mr. Serafin's seeding/editing practice....When he is working on his files instead of joining all the files together into a larger file, "a complete set" as it is, he works on each track on it's own. Now here is the problem with doing that...When applying the "digital work" to the files individually, like: EQ, noise reduction, Time Smear correction, changing the sample rate/word length, and dithering after the downsample...you have added, removed, altered the information "in" the file. The tracks when they first get transfered to his DAW may be correctly residing on sector boundaries, but after processing each track in this mannor the changes in the data will cause either addition or subtraction of the data in the files, therefore pushing the data outside the sector boundaries. IF one were to join all the tracks into a "set" as it were, this problem is completely avoidable. The processing is moving the data in the "whole" file this way, not in each track. Then after you have done your sound processing, if you were to retrack the show, you would find that the sector boundaries would more than likely be ok (not taking into account the other factors such as overall file length, etc...) So the moral of the story here is this: If you are going to edit a show in any way, join all the files that belong as a Set together before you do any editing, after editing retrack the show, and lastly run the freshly retracked show through shntool to verify your cuts are correct. When seeding, if you are going to perform any sound processing, make sure you do it to the untracked file...then track the set. This will save you all kinds of headaches in the long run, and it will keep you from having people rip your methods apart for causing such an easily avoidable error such as Mr. Serafin constantly does.

See also: SeedingGuidelines