robert taylor said:

"--nogap is the same as making one big mp3 file with the encoder and then splitting it youself on mp3 frame bounderies. --nogap has one extra nicety in that it will reset the bitreservoir at track bounderies, so the first frame of all the mp3 files will be valid when played back by itself.

For nogap to work in a player with a play list of the 10 mp3 files, the player just needs to not reset the encoding engine between each mp3 file. (and hope that the sample rate and number of channels is not changing from one mp3 file to the next). If you add tags (id3 or vbr header) the decoder also has to have the logic to look for and remove these tags when going from one mp3 file to the next. So a simple "gapless plugin" which just sends the playlist to the decoder as a single mp3 stream will decode the vbr headers as 20ms of silence."

In reply to:

---

A patch was recently committed to fix --nogap in the most recent lame alpha builds. I'll give it a shot tomorrow and report on how it turns out with 2.0.

---

Oh, now this would be great! If that works as expected, then gapless playback will work for sure as long as you make sure that the .WAV files are multiples of the MP3 frame length.

Michael Grant
12GB Green
080000266

In reply to:

---

Remember that the player has got NO CLUE that Track B is a continuation of Track A, and it's actually doing the Right Thing to dump the bit reservior and restart the decoder when it encounters a new track. If it didn't, then two disparate tracks would have a glitch at the boundary, and that's not good either. In fact, it would be worse. So I think the player is doing the Right Thing now. It's certainly doing the best it can with the material it's been given.

---

Yeah, you're right, leaving the bit reservoir in place could cause some sonic issues with normal tracks. But I figured that if the track isn't gapless to begin with, that it would be even more inaudible with "gapped" tracks than with "gapless" ones. I could be wrong about that though.

It might be nice to have a configurable setting to preserve the bit reservoir or not. This setting could be made global, or it could be set for each playlist.

Still, Terminator says he's looking into the '--nogap' setting on LAME, which should be a real help.


Michael Grant
12GB Green
080000266

In reply to:

---

"--nogap is the same as making one big mp3 file with the encoder and then splitting it youself on mp3 frame bounderies. --nogap has one extra nicety in that it will reset the bitreservoir at track bounderies, so the first frame of all the mp3 files will be valid when played back by itself.

---

Ah, this looks like just what I was hoping for. But I'm a bit concerned about this:

In reply to:

---

For nogap to work in a player with a play list of the 10 mp3 files, the player just needs to not reset the encoding engine between each mp3 file.

---

Does he mean to say the "decoding" engine? And why can't the engine be reset, if the bit reservoir is no longer an issue?

Michael Grant
12GB Green
080000266

In reply to:

---

If they do, then it would suffice never to reset the decoder: files made by cutting a large MP3 would play exactly as one, and in the case of independend files the 'leftover' from the first would not spoil the beginning of the next.

---

This is actually the behavior I'm hoping for... but the last part about independent filies is not 100% certain. Frankly I haven't heard evidence that the leftover bit reservoir information would be audible during the gap silence. But someone else may have heard it. And just because I'm more sensitive to glitches in gapless playback doesn't mean I should irritate someone's sensitivity to noise in a gap :-)

I think the absolute best solution would be to have a flag in each playlist that, when enabled, prevents the decoder from returning to neutral between consecutive tracks.You could use a similar playlist flag to tell the decoder to fade in and/or out gapless tracks when they are played independently (e.g. during shuffleplay).

I hope that this discussion is not suggesting that I'm unappreciative of the efforts to get gapless playback working. Frankly I'm thrilled about what's gone into 2.0, including what already has been done on gapless playback. In fact, I am now encouraged to keep my empeg; the announcement of its impending demise really had be thinking about selling it.

But I do think that the beta period is a perfect time to be making this noise, so I'm doing so!

Michael Grant
12GB Green
080000266

Here's another phenomenon I've encountered on one of my CDs that I'm trying to understand how best to duplicate on an EMPEG.

If you've created CUE sheets with Exact Audio Copy (for example), then you've seen that most CD tracks have an "Index 0" timestamp and an "Index 1" timestamp. Usually, the time period between Index 0 and Index 1 contains the inter-track silence, and the music proper begins at Index 1. If you use the "skip forward" or "skip backward" buttons on a CD player, it will always seek out the Index 1 timestamp.

However, there is nothing that prevents the CD from having sound between Index 0 and Index 1. Indeed, one of my CDs does just that. In this case, what happens is that the band starts playing, and then decides to stop playing and start over again. The "false" start lies between Index 0 and Index 1, and the second "true" start actual track begins at Index 1. There's no gap in between. If you try to hit your CD player's "skip backward" button to go back to the beginning of the track, you won't hear the false start; it will just move back to Index 1 and continue from there.

Now when I'm encoding this track, I have two choices:

1) Keep this false start as part of the MP3 file.
2) Split the track into two pieces between the false start and the true start, and encode the two pieces in some sort of gapless fashion.

The first method is the easiest. The second has the potential to better emulate the behavior of the CD when I'm playing it back. However, the one quirk of the second method is that I wouldn't want the "false start" to play if I was in "shuffle play" mode. So, I would need to find a way to ignore that track if I'm shuffling, but I'd want it to play if I'm going sequentally through the album.

What do you think, folks?


Michael Grant
12GB Green
080000266

Greetings!

Having actually put these "hidden audio" bits in some of my past projects, you have to be careful when ripping. Many rippers will add this onto the end of the previous track. Not a problem if the audio is before track 1, but it can be annoying if it is elsewhere in the album.

I personally prefer to rip it as its own track, and only have it in linear playlists. I omit it from anything I do that is randomized. It is the only way I know of to give the audio clip its own identity, independent of the tracks before and after it.

Paul G.
SN# 090000587 (96GB Smoke)
SN# 030103046 (10GB Blue - Emergency Spare)

What do you think, folks?

Start a new thread on this subject! I hate it when the subject is changed somewhere in the thread...

­­______________
Edwin de Vaan

In reply to:

---

Start a new thread on this subject! I hate it when the subject is changed somewhere in the thread...

---

Fair enough. I think it was reasonable to put the gapless stuff here, but this particular subject change is a bit too far off-topic. I may just do that later today...

Michael Grant
12GB Green
080000266

I think it was reasonable to put the gapless stuff here

Of course! It is always useful to put up your thoughts on this BBS. I sincerely hope a more gapless playback is possible with version 2.0. Let's wait and see!

­­______________
Edwin de Vaan

In reply to:

---

Beta 1 will be released to the alpha test team tonight, and if they don't break it horribly it will go public on Monday. If they do break it horribly, we might release their home addresses to the BBS *grin*

---

Ok I think its about time. Whats their home address? or just thier IP addresses will do!!


-Doug
Mk2-12G Blue (Now A/R Green)

Hope Rob isn't regretting putting a time on a release!

Honest Rob, all is forgiven!!

as long as v2 is released soon!

Murray
Go-Wit-Da-Fro

Monday is quickly fading here on the west coast! 30 minutes left.


|| loren.cox ||

20 minutes and counting...
-mark

This space used in memory of the victims from 09/11/01

Maybe he meant for the Alpha testers to release it for him... and that's why he was going to hand out their addresses if it hadn't appeared by Monday...

*grin*

Kim

Ah, but there is a guy in an outrigger canoe somewhere in the middle of the Pacific with a large satellite dish and and laptop who still has 4 hours of Monday left...

__
Unit serial number 47 (was 330 in the queue)...

Help me I'm HUNGRY .......................

Argh !!!!!!!!!!!!!

Empeg II - 10 Gb - Red Fascia - 750 Songs -

- I Will Strike From the Grey -

looks like this thread's replies is gonna beat the EOL one :)
not sure about # views tho..

That will happen in 2.x but not the first release. The plan is to go back to semi-regular releases with functionality updates in each.

Rob

WOW! You're back! Can you enlighten us with more details on the release time/date of v2.0?

­­______________
Edwin de Vaan

Anyone want to hazard a guess?

Has to be the improved gapless playback. Right?

tanstaafl.

"There Ain't No Such Thing As A Free Lunch"

Right. You know me too well.

___________
Tony Fabris