#12347 - 31/07/2000 18:48
Idea for gapless album playback
|
journeyman
Registered: 28/04/2000
Posts: 84
Loc: Menlo Park, CA USA
|
Hi folks, I've been thinking a lot recently about the gapless playback issue. Now don't worry, I've searched the posts and so forth, so I know it's been bandied about before. I know from experience and theory that creating a set of gapless MP3 files can be tedious and painful---particularly if, as happens sometimes, the track boundaries defined by the source CD are not exactly correct and require manual editing. Basically, here's my idea. It seems to me that the only way to insure perfect fidelity gapless playback of a single album is to encode the entire album as a single MP3 file. That's great, and of course empeg will have no problem with that if all I wanted to do was play the entire album. But DAE programs like Exact Audio Copy and CDRWin can create "cue sheets" that go along with a full-album extraction that documents the gaps and track boundaries. (One thing that's nice about this is that, if you disagree with the track boundaries that the CD master chooses, you can just adjust the cue sheet instead of re-ripping the offending tracks.) Certainly a system as flexible as the empeg could be configured to use these cue sheets in an intelligent way. What I envison is that, when presented with an album-size MP3 plus a cue sheet, the empeg would create a set of "virtual tracks" that for all intents and purposes function like normal tracks. When the user selects an individual track, it will know to read the cue sheet for the track position, then seek along the MP3 file to find it. Of course, when playing the entire album, it need not do any seeking, and of course it will sound like a continuous stream of music as intended. I'm certainly not the only person to think of this; in fact, seeing someone else's version of this idea inspired me to post here. And of course, the program CDRWin has been a popular tool for some time for, as they call it, "disk-at-once" (DAO) recording using cue sheets. Comments? Ideas? Michael on the list for a mk2...
_________________________
Michael Grant
12GB Green
080000266
|
Top
|
|
|
|
#12348 - 31/07/2000 23:10
Re: Idea for gapless album playback
[Re: mcgrant]
|
veteran
Registered: 16/06/1999
Posts: 1222
Loc: San Francisco, CA
|
In reply to:
Poster: mcgrant Subject: Re: Idea for gapless album playback
Hi folks,
I've been thinking a lot recently about the gapless playback issue. Now don't worry, I've searched the posts and so forth, so I know it's been bandied about before. I know from experience and theory that creating a set of gapless MP3 files can be tedious and painful---particularly if, as happens sometimes, the track boundaries defined by the source CD are not exactly correct and require manual editing.
Basically, here's my idea. It seems to me that the only way to insure perfect fidelity gapless playback of a single album is to encode the entire album as a single MP3 file. That's great, and of course empeg will have no problem with that if all I wanted to do was play the entire album.
But DAE programs like Exact Audio Copy and CDRWin can create "cue sheets" that go along with a full-album extraction that documents the gaps and track boundaries. (One thing that's nice about this is that, if you disagree with the track boundaries that the CD master chooses, you can just adjust the cue sheet instead of re-ripping the offending tracks.)
Certainly a system as flexible as the empeg could be configured to use these cue sheets in an intelligent way. What I envison is that, when presented with an album-size MP3 plus a cue sheet, the empeg would create a set of "virtual tracks" that for all intents and purposes function like normal tracks. When the user selects an individual track, it will know to read the cue sheet for the track position, then seek along the MP3 file to find it. Of course, when playing the entire album, it need not do any seeking, and of course it will sound like a continuous stream of music as intended.
A very good idea that has been suggested before, although your post was much more detailed/helpful:) It would help those of us with bootlegs as well - alot of my music is live recordings which don't really have any track limits. I've got god-knows-how-many track names such as "simons 6-4-99 - track1" --> track21:)
or how about pway track 1 ---> pway track 499
A cue sheet would be the answer to long recordings as well.. -mark
...proud to have owned one of the first Mark I units
|
Top
|
|
|
|
#12349 - 01/08/2000 00:45
Re: Idea for gapless album playback
[Re: mcgrant]
|
carpal tunnel
Registered: 20/12/1999
Posts: 31600
Loc: Seattle, WA
|
Actually, you wouldn't need a cue sheet at all.
You'd just need a ripper program that was smart enough to:
a) Encode the entire album as one long MP3, then
b) Split the big MP3 cleanly at exact frames that are close (within .026 second) of where the CD thinks the track boundaries should fall.
Then everything would play back perfectly in player programs that cached the stuff properly.
I've just never seen a program that would do this for you.
See, the problem isn't the player's fault, it's really the encoder's fault. The problem is the encoders inserting silence into the first and last frames. If the ripper did one huge "encode" and then split it after encoding, it would work around the encoder's problem.
The real solution would be to go back in time and locate the folks designing the first MP3 specification, and force them at gunpoint to include support for continuous-track material (such as allowing a frame's length to be specified so that the last frame wouldn't have to be padded, and then allowing the first frame to start on the first sample).
Does anyone know the technical reason that the first frame gets padded with silence? I know why the last one does, but why the first? One of the Fraunhofer techs once told me that it was the nature of the algorithm and there was nothing that could be done about it, but I'm curious as to why.
___________ Tony Fabris
|
Top
|
|
|
|
#12350 - 01/08/2000 01:14
Re: Idea for gapless album playback
[Re: tfabris]
|
journeyman
Registered: 28/04/2000
Posts: 84
Loc: Menlo Park, CA USA
|
Actually, you wouldn't need a cue sheet at all. You'd just need a ripper program that was smart enough to: a) Encode the entire album as one long MP3, then b) Split the big MP3 cleanly at exact frames that are close (within .026 second) of where the CD thinks the track boundaries should fall. Then everything would play back perfectly in player programs that cached the stuff properly.There is indeed software to do this, it's called musiCutter. The concept is discussed in more depth on the "live-mix" section of the r3mix site. However, this latter site highlights some of the problems with this technique, too. One of the is the obvious: the need for a truly gapless player. But another reason that it mentioned is ultimately the reason I decided that it was likely best not to split the big MP3 files into little ones. And that is that, according to the r3mix site and other information I had encountered previously, it is not necessarily true that all the information required to reproduce a single frame of sound is contained within the frame itself. In fact, it can spill over into adjoining frames. Perhaps someone with more exact knowledge of the MP3 format and encoding process can help with this (the "byte reservoir?"). But if this is indeed the case, then the gapless player would not be able to achieve truly perfect reproducton without some sort of knowledge that the consecutive MP3 files contain information about each other. Without that information, the reproduction might certainly be acceptable, but not perfect.
_________________________
Michael Grant
12GB Green
080000266
|
Top
|
|
|
|
#12351 - 01/08/2000 17:31
Re: Idea for gapless album playback
[Re: tfabris]
|
member
Registered: 16/12/1999
Posts: 188
Loc: Melbourne, Australia
|
Actually, you wouldn't need a cue sheet at all.
Hmm. Personally, I quite like the idea of removing the one to one correspondence between a file and a tune. Although, I guess playlists are already like a virtual tune, that consists of lots of files, if you never shuffle that playlist. But it would be a lot easier to tweak some kind of metadata than to move frames from one file to the other, if you change your mind about where the boundaries should be.
Also, you could have some overlap, so that (for instance) track 2 begins before track 1 ends. Your cue sheet could contain some volume information as well, so that you can fade in at the beginning, and out at the end. This might let you still listen to sections of a continuously mixed album in shuffle mode without that jarring effect.
Richard.
|
Top
|
|
|
|
#12352 - 02/08/2000 03:30
Re: Idea for gapless album playback
[Re: mcgrant]
|
pooh-bah
Registered: 13/09/1999
Posts: 2401
Loc: Croatia
|
I certanly second your cue-sheet idea. (Actually, I think that I already sugested something like that - but not in such an articulate way - or at least thought about suggesting .) As you say, it would be nice to have cue-sheet indices into long pieces irrespective of their utility for gapless album playing (e.g. different parts of a single movement of a symphony, or 'loud part' of Locomotive Breath :) So, any comments, guys@empeg? Cheers! Dragi "Bonzi" Raos Zagreb, Croatia #5196
_________________________
Dragi "Bonzi" Raos
Q#5196
MkII #080000376, 18GB green
MkIIa #040103247, 60GB blue
|
Top
|
|
|
|
#12353 - 26/08/2000 00:06
Re: Idea for gapless album playback
[Re: mcgrant]
|
journeyman
Registered: 28/04/2000
Posts: 84
Loc: Menlo Park, CA USA
|
I was just perusing through the technical documentation for the Personal Jukebox (you know, that PJB-100, hard-drive based portable MP3 player); Compaq released all sorts of info about the USB and hard drive interface. It's available for download here if you're interested. Anyway, I point this out to quote the following from the documentation: "One feature of the Personal Jukebox is its ability to store an entire CD as a single MP3 bitstream, while still allowing individual access to each track. The way this is accomplished is to store the CD as a continuous chain of blocks, but create Track entries for each track pointing into the allocated chain of blocks." Hey, that sounds familiar! Very, very similar to what we were talking about in this thread. Of course I don't believe the windows jukebox management software can handle this feature at the moment, but the API can... Michael
_________________________
Michael Grant
12GB Green
080000266
|
Top
|
|
|
|
#12354 - 26/08/2000 00:09
Re: Idea for gapless album playback
[Re: mcgrant]
|
veteran
Registered: 16/06/1999
Posts: 1222
Loc: San Francisco, CA
|
In reply to:
"One feature of the Personal Jukebox is its ability to store an entire CD as a single MP3 bitstream, while still allowing individual access to each track. The way this is accomplished is to store the CD as a continuous chain of blocks, but create Track entries for each track pointing into the allocated chain of blocks."
Great; knowing compaq, they probably put a patent on something we've been talking about all along...
...proud to have owned one of the first Mark I units
|
Top
|
|
|
|
#12355 - 26/08/2000 06:06
Re: Idea for gapless album playback
[Re: dionysus]
|
pooh-bah
Registered: 13/09/1999
Posts: 2401
Loc: Croatia
|
Great; knowing compaq, they probably put a patent on something we've been talking about all along...We will all testify the idea is obvious (non-obviousness - for people in the field, not patent clerks - being prerequisite for a patent). Also, does discussion here - this being a public forum - constitute a disclosure? Are those cue sheets previous art? Finaly, are we sure that the PJB software have not actually been commisioned by Compaq from a certain small high-tech company that is going to surprise its own customers with something similar? Cheers! Dragi "Bonzi" Raos Zagreb, Croatia Q#5196, MkII#80000376, 18GB green
_________________________
Dragi "Bonzi" Raos
Q#5196
MkII #080000376, 18GB green
MkIIa #040103247, 60GB blue
|
Top
|
|
|
|
#12356 - 27/08/2000 08:11
Re: Idea for gapless album playback
[Re: bonzi]
|
new poster
Registered: 27/08/2000
Posts: 1
|
hello, i'm new here and maybe this has been asked/covered before, but does the empeg have gapless playback? for example if i have a dj mixed set will it go from one song to the other seamlessly (like the gapless output plug in for winamp) or is there a tiny gap or pause? this is the only thing stopping me from putting myself on the waiting list :( any info would be helpful. thanks for your help!!
|
Top
|
|
|
|
#12357 - 27/08/2000 10:30
Re: Idea for gapless album playback
[Re: vegas2000]
|
addict
Registered: 15/07/1999
Posts: 568
Loc: Meije, Netherlands
|
maybe this has been asked/covered beforeAt length. One of the places where it was discussed is this one Henno mk2 6 nr 6
_________________________
Henno
mk2 [orange]6 [/orange]nr 6
|
Top
|
|
|
|
#12358 - 27/08/2000 12:57
Re: Idea for gapless album playback
[Re: rjlov]
|
enthusiast
Registered: 17/08/2000
Posts: 334
Loc: Seattle, WA. USA
|
I've stumbled around the issue of live recordings and MP3's for a while. I have not expended a lot of brain power on it, but some of the ideas here make me want to look at it again.
I hate the little pauses "gap-less-less" playback creates, but I also don't like the harsh transitions that occur when a live track gets mixed into a random playlist. (It also bugs me when the intro to the next song gets tagged to the end of the previous song... but what are you going to do?) My solution to this has been to record one full MP3 and several track based MP3s, with my own fade-in and fade-outs. Really messy...
I'm not a big fan of crossfading, but that would have some interesting uses... I used to have a radio show and one of my gimmicks was the "dream concert." I would play a series of Live songs by different artists. If I was quick enough to crossfade the crowd noise just right, it would sound like they were all on stage together... Pretty fun. If I was really clever, I remembered to have a large gap of crowd noise (usually found in the first or last track in a live album) to toss in the station id. Made it sound like I was MCing...
Anyway, I'm just thinking out loud... This BBS has made me rethink a lot of my organization plans.
_________________________
Brian H. Johnson MK2 36GB Blue, currently on life support "RIP RCR..."
|
Top
|
|
|
|
#12359 - 27/08/2000 17:03
Re: Idea for gapless album playback
[Re: bootsy]
|
carpal tunnel
Registered: 20/12/1999
Posts: 31600
Loc: Seattle, WA
|
My solution to this has been to record one full MP3 and several track based MP3s, with my own fade-in and fade-outs. Really messy...
That's not messy at all, that's a great way to do it. And MP3Trim has just added the ability to create a fade in/out effect directly upon an MP3 file, so the process of doing this is fairly painless now.
___________ Tony Fabris
|
Top
|
|
|
|
#12360 - 27/08/2000 17:08
Re: Idea for gapless album playback
[Re: vegas2000]
|
carpal tunnel
Registered: 20/12/1999
Posts: 31600
Loc: Seattle, WA
|
Recent revisions of the Empeg software have significantly improved its ability to play tracks back-to-back without gaps. I've got some stuff where I can't hear a gap now between continuous tracks. Certain other tracks where the gap occurs over a smooth continuous note are a little more obvious but very tolerable. Here is a FAQ entry which quotes some of our discussions on the subject. ___________ Tony Fabris
|
Top
|
|
|
|
#12361 - 01/09/2000 23:16
Re: Idea for gapless album playback
[Re: tfabris]
|
journeyman
Registered: 28/04/2000
Posts: 84
Loc: Menlo Park, CA USA
|
a) Encode the entire album as one long MP3, then b) Split the big MP3 cleanly at exact frames that are close (within .026 second) of where the CD thinks the track boundaries should fall.
Sorry to resurrect this thread, but I just tried this. I encoded Stevie Ray Vaughan & Double Trouble's "Live at Carnegie Hall" as a single MP3 file, and then I used musiCutter to split the file into its parts.
When the empeg plays it back, it inserts small but audible gaps between the tracks. Of course, I don't think the empeg is necessarily trying to go gapless here. If it's not, then it sure seems like it could if it just preserved the MPEG decoder state between files.
Oh well, for now it sounds as good as I'm gonna get I guess.
Michael Grant 12GB Green 080000266
_________________________
Michael Grant
12GB Green
080000266
|
Top
|
|
|
|
#12362 - 01/09/2000 23:21
Re: Idea for gapless album playback
[Re: mcgrant]
|
carpal tunnel
Registered: 20/12/1999
Posts: 31600
Loc: Seattle, WA
|
Question: The resulting SRV files... do you have a PC-based gapless player or plug-in that will play those tracks seamlessly?
___________ Tony Fabris
|
Top
|
|
|
|
#12363 - 01/09/2000 23:49
Re: Idea for gapless album playback
[Re: tfabris]
|
journeyman
Registered: 28/04/2000
Posts: 84
Loc: Menlo Park, CA USA
|
Question: The resulting SRV files... do you have a PC-based gapless player or plug-in that will play those tracks seamlessly?
Now that was a very good question. I hadn't checked, because I've already gotten into the habit of using the empeg as my normal playback device :-)
The gapless playback device Winamp supplies can't handle them, no. I'll look into others.
Michael Grant 12GB Green 080000266
_________________________
Michael Grant
12GB Green
080000266
|
Top
|
|
|
|
#12364 - 04/09/2000 22:32
Re: Idea for gapless album playback
[Re: tfabris]
|
journeyman
Registered: 28/04/2000
Posts: 84
Loc: Menlo Park, CA USA
|
More rambling about gapless playback...
Question: The resulting SRV files... do you have a PC-based gapless player or plug-in that will play those tracks seamlessly?
After some experimentation, I find that the latest WinAmp, coupled with the "Dudsoft Gapless Output" plug-in, does the best job I've heard so far at removing gaps from my "split" MP3 files---that is, my live SRV album encoded as a single MP3 file, and split into multiple files across track boundaries.
It's very satisfying to hear, after getting used to the inevitable gap. Still, it's not perfect. Between some tracks I cannot hear the gap, and between others I can hear the faintest blip. So, I suspect that the "bit reservoir" is being flushed between tracks---and its effect on the sound depends on the amount of information saved there.
One thing I did differently this time is to feed WinAmp files with absolutely no metadata; i.e., no tag data, nothing but pure audio frames.
It may seem like this is my quixotic crusade here, but after some consideration I feel like the "split MP3" technique that Tony proposed awhile back is definitely possible. The conditions that it places on the MP3 decoder seem simple enough: 1) the decoder must be able to accept multiple MP3 files in sequence without introducing a gap of its own; and 2) (importantly) it must preserve all of its "state" information between files.
It is factor #2 that I don't think WinAmp is quite getting right. And yet, I don't believe #2 will affect normal MP3 files; my examination of LAME mp3 files is that they seem to "flush out" the bit reservoir at the end of the track by adding blank frames at the end as necessary. So, such a file would not corrupt the beginning of the next file.
Michael Grant 12GB Green 080000266
_________________________
Michael Grant
12GB Green
080000266
|
Top
|
|
|
|
|
|