I have trouble with the transition between two specific tracks
CD: Dj Convention - Code Eleven (Disc 2)
Tracks 1 and 2
Ripped with Plextor 24/10/40a + eac (read offset +98)

A short silence is introduced between the tracks when encoded with 1.01g and played back with the 0.92g winamp plugin. The wavs play back fine without gaps. in_mpc v0.92d plays back the mpcs without gaps. The other tracks of the cd play gapless with 0.92g. When decoding the mpcs with mppdec 1.01g the files have the exact same file size as the originals and play back without gaps. It seems there is a bug in the 0.92g winamp plugin.

The wavs have a sample length of:
1: 11327330 samples (45309364 bytes)
2: 12900720 samples (51602924 bytes)

I provided the wav file size in case I made errors reading the sample lengths. If there is need, I could provide the complete wavs (I don't have web space but I could upload them to an ftp) short samples won't do. If you need further details, please ask. Hope this helps to sort out the gap problem!

Download my fixed version here: http://www.naivesoftware.com/mpcfix/in_mpc.zip

I've sent the necessary changes to Frank, so hopefully the next version will be fixed.

The encoded files now play back without any noticable gap. The problem seems fixed!

Thank you very much for the quick and helpful response!

I updated my encoder, decoder, and winamp plugin to the new versions, but I'm still having gap problems.

The gaps that I had with the "d" version are still formed when I use the "g" version. Most notably the Narayan/Firestarter transition discussed in the "Is gapless MPC encoding possible?" thread. The wavs are perfect, but there is a very obvious click between the mpc's. I can post samples if that will help.

I don't know what is wrong, but the only difference I observed while using the "g" versions is the playback of 2 tracks [Night/Hide Nowhere from Biomech by Devin Townsend] that were irritating me previously, because there was a tiny pop between them. With the Winamp plugin 0.92g, the playback of the mpc's is now perfect. I've tried Trelane's version of the new plugin also, and what I've noticed is that the little pop is audible with Trelane's plugin and not audible with the original 0.92g plugin.

I am also having problems with another CD - The Perfect Element I by Pain Of Salvation. There is a rather loud pop in my mpc files for Falling/The Perfect Element.

Since I am having problems with so many files, I can't help but think that I am doing something wrong, as the new plugin will produce errors (theoretically) on only about 1% of encoded files. The problem is - I have no idea what I am doing wrong!

I am currently using:
mppenc 1.01g
mppdec 1.01g
Musepack plugin for Winamp 0.92g
EAC for ripping to WAV, MPC Batch Encoder for encoding to mpc.

My CD-ROM drive may be a factor, as in the other thread, PoisonDan had different results with his files after replacing his CD-ROM drive. How much can this affect the process?

Arrghh! :mad:

Thanks for any and all help!

JS

JS,

Have you tried encoding my WAV files again with the new encoder ? And did you try listening again to my MPC files with Trelane's new WinAmp plugin ?

I must say that, unlike you, I didn't hear any click at all in my MPC files. I suspected it might have been a problem with the WinAmp plugin (I don't use WinAmp, I use Media Jukebox).

I also decoded my MPC files back to WAV files and burned them to CD-R to make sure it's not a player issue. The CD-R also sounded perfect on my stereo.

After you have listened to my samples again (encoded by you and encoded by me), could you let me know what the result is ?

If you still hear a click, I'll install WinAmp with the same plugin as you and check if I can hear it too.

JS:
I should mention that my version is just a quick little hack of the official Winamp plugin. I only had a limited number of samples and time to test it with. Everything sounded good on this end. Make sure you don't have Replay Gain enabled on a title-based setting, as this could potentially produce a click. If clicks still appear, try encoding the files with the latest encoder.

PoisonDan:
The MJ plugin should be perfect, as I've spent significantly more time (all day, everyday for the past two months) testing it. If you find any problems, let me know. Thanks

JS:
I have the Narayan/Firestarter samples that were posted to test with... would you mind making short clips of Devin Townshed and Pain Of Salvation so that I can test those as well?

Thanks

It IS perfect, and all your hard work is much appreciated.

How can short clips help, if the main problem has to do with the total sample length of a track? I guess what would help most (if webspace/speed is the limiting factor), is posting the amount of samples. I will try those prodigy tracks later on allthough I'm not sure if your drive will create wavs of the same sample length as mine.

Last 5 seconds of the first track, first 5 seconds of the second track in the pair of offending samples. This suffices for the purpose of testing for gapless playback as this is where the problem lies.

Let's talk only about one single track. Suppose this Track has 100 samples. The encoder only produces errors with sample lengths that are a multiple of 10. Now I only provide the last 25 samples. These samples would then encode/decode error free. I may have understood the thing wrong, why mpc has these gap problems but if I'm not mistaken, it had to do with the total sample length. Not the signal is the problem but it's length. :confused:

Allthough since you were able to fix my gap bug and code the MJ mpc plugin (?), I assume you know alot more about the decoding process than I do.

You are correct in your thinking... I didn't think about it that way (I guess when JS posted the previous shortened samples, we just got lucky ). I will generate some files based on Frank's criteria for problem samples for gapless encoding.

7 seconds of PCM is a case where the bug occures.

The issue with the sample lengths that Gecko brought up had not occurred to me. The gap problems are still present though, whether I truncate the wavs or not.

If the entire wav file is necessary for investigating the problem, I am fine with it, as I can just run an http/ftp server off my computer. But I wonder who will be willing to download these huge 50mb wavs!

PoisonDan:
I tested the files that you put up originally. Same as before, the huge gap is gone but there is a tiny pop still present (Note: it is VERY subtle; you will have to listen carefully). The results with your mpc's and the mpc's I encoded from your wav's are the same - that tiny pop.

This is in Winamp. I think I will grab myself a copy of Media Jukebox and see if I hear the same gaps.

Trelane, I can upload the other files with no problem, but should I truncate them? I think, going by Gecko's information, that I should upload the entire wav's, but I think it will be a problem for you to download those huge files. I am fine either way, though.

JS

If you are noticing the click on the truncated files, just post those. If not, don't worry about it.

Question: have you tried playing the files with mppdec? I can't detect a pop with either the MJ plugin or mppdec (the decoding kernels are similar). Did you reencode the files with the latest encoder? You can test out the files with mppdec like so: mppdec file1.mpc file2.mpc

Sorry Trelane that I took so long to reply. I've updated my encoder, decoder and output plugin with the "h" versions, and I've noted my observations below.

I've uploaded the problem files.

The Devin Townsend files:
http://hammer.prohosting.com/~jys02/03Night.wav
http://hammer.prohosting.com/~jys02/04HideNowhere.wav

http://hammer.prohosting.com/~jys02/03Night.mpc
http://hammer.prohosting.com/~jys02/04HideNowhere.mpc

When I listen to the mpc's in Winamp, I hear a very tiny pop between the tracks. When I listen using mppdec, the playback is flawless. What do I need to do to get Winamp to play like this?



The Pain Of Salvation files:
http://hammer.prohosting.com/~jys02/11Falling.wav
http://hammer.prohosting.com/~jys02/12ThePerfectElement.wav

http://hammer.prohosting.com/~jys02/11Falling.mpc
http://hammer.prohosting.com/~jys02/12ThePerfectElement.mpc

Between the mpc's, there is a very loud pop (in Winamp). Played with mppdec, the loud pop is gone, but there is still a small pop audible.



What is causing the mpc's to play differently in Winamp and mppdec?

Thanks for helping me out!


JS

edit: links fixed

mppdec is Frank's decoder. It is newer. The Winamp plugin is based off of Andree's decoder, which is quite dated these days.

I get a 404 on those files, by the way.

Sorry, I didn't know my server was so anal about file capitalization... fixed now

So the Winamp mpc plugin (the new h-version) is based on Andree's old decoder? How would I make it so that I can use Frank's new one? Or did I not understand you correctly..?

Thanks,

JS

It's based on the source code of Andree's decoder... so one would have to rewrite the Winamp plugin to base it from Frank's.

This is not a gap bug, but an artifact. A quick test using the braindead profile and the click disappeared. I figured it couldn't be a gap bug because it was only occuring in the right channel.

My recommendation: encode these two with braindead or mess around with the settings manually.

Good luck.

--xtreme (or --minval 2) should be enough.

Yup Trelane, I tried encoding the wavs using the braindead profile and that little gap between the Pain Of Salvation files is gone. (using mppdec)

So, as far as I understand, since the Winamp mpc plugin is based on Andree's old code, I wouldn't be able to get files to play as flawlessly in Winamp as they do with mppdec (unless a new plugin is written based on Frank's new decoder)?

Using Media Jukebox, the playback should be flawless, right? (With your mpc plugin )

I've tried using MJ, but the size of the program turned me off initially. I just want to use a simple program to play back my audio files, not some huge, elaborate one, which MJ seems to be.

But I may just convert to MJ because of this decoding issue.

Any other advantages MJ has to Winamp?

(I've heard that it takes up CPU and slows the system down even after uninstalling because it adds a ton of junk to the registry... What is that about?)

Thanks Trelane and Frank!

JS

I tried your files too and Winamp plugin (version 0.92k) decoded files almost identically with mppdec.exe. Only least significant bit was different. The gap was present also with files decoded with mppdec. And the gap was gone with --xtreme, just as Frank suggested.
Also, 03night ends in silence, I don't hear anything wrong with it.

Playback is flawless in the Winamp plugin as well.

One thing I noticed, Xtreme isn't enough... didn't try minval, though.

Sorry, my fault, I had the Winamp EQ on. The braindead files play perfectly with 0.92k, EQ off.

However the --xtreme files (falling & perfect element) still have a little pop in the right channel.

Do you guys use the Shibatch EQ plugin for Winamp? I just made do with the Winamp EQ instead because the Shibatch EQ messed up my visualization and playback when I manually switched tracks.

JS

Note: The mpc's I put up in this thread were encoded using the --xtreme profile, if that clears up anything.

A few words about Media Jukebox.

It is by far the best media player/organizer around. Of course, you can't really compare it with WinAmp. WinAmp is a small extensible player, while MJ is a big all-in-one utility: media player, organizer, burner, ripper (even secure), cover editor, wave editor, media converter, tagger, media server and lots more.

It plays about all audio and video formats you could wish for (about 60 formats supported), had built-in support for replay gain (all audio formats), equaliser, DSP, gapless output, crossfading, etc. etc.

It is also very customizable with skins, smartlists, visualization studio, HTML track info pages, etc.

And one more very important point: it has a very active message board, where the developers also hang out, and they really listen and act promptly to the feedback, feature requests and bug reports of their users. Their involvement with their users is probably one of the main reasons why MJ is so great.

About the only negative point about MJ is the fact that the vast amount of features may somewhat intimidate novice users: there's just so much to explore here.

About the performance: as far as I can tell, CPU usage for playing is about the same as WinAmp (and definately better than e.g. Windows Media Player 7.x). Of course, the visualizations can take quite a bit of CPU cycles, but MJ only uses "idle" CPU cycles for the visualizations. This means that the visualizations will never slow down the system (even though some users might suspect it will, when they look at the CPU usage). And, if some other CPU intensive task is running in the background, the visualizations will simply become more choppy, but the system will be working perfectly otherwise.

About this "registry junk": I'm not aware of this, but of course MJ 8.0 has been in beta for quite some time now, and maybe this was present on earlier beta versions. You should never judge an application by bugs in one of the beta versions.

If you want to check it out, I suggest you download the latest beta version of MJ 8.0 from the forum (Interact -> Talk back). The latest beta versions appear to be pretty solid.

In case you are wondering: no, I don't get paid by J River

You and Trelane are right, it is still present with xtreme. I had too low volume to hear it. It's probably caused by the hack nature of this fix.


I never use any EQ.