Gap problem in MPC Options

[shadowking]

A 'blip' is heard in the change from track 1 to 2. The album is 'the kick inside' by Kate Bush. Problem affects foobar 8.2, winamp 5.x, mppenc 1.14 / 1.15r q5-10. No problem with mp3 and vorbis.

Can other please confirm.


thanks.


1st attachment = t1.flac

2nd = t2.flac

[Mr_Rabid_Teddybe...]

Yes. I got a minor 'click' between the tracks thats not in the original flacs using mppenc 1.14b, "--xlevel --quality 5 - %d", fb2k 0.8.2, XPsp1.

Won't bother to check more settings/progs now, since I'm going off to bed, prolly others will help you test more exhaustively --- but if this is for real it's a minor blow to musepack, since it's known as one of the "naturally gapless" codecs......

[Sebastian Mares]

Pio2001 also reported this behavior in the following thread: http://www.hydrogenaudio.org/forums/index....ndpost&p=179277

LAME, MusePack, AAC and WMA Standard seem to suffer from this problem, while Vorbis and WMA Professional not.

This post has been edited by Sebastian Mares: May 27 2004, 08:29

[Mr_Rabid_Teddybe...]

Pio2001 also reported this behavior in the following thread: http://www.hydrogenaudio.org/forums/index....ndpost&p=179277

LAME, MusePack, AAC and WMA Standard seem to suffer from this problem, while Vorbis and WMA Professional not.

Sorry. Now also tested with:

FAAC (1.24+) "-w -q 100 -c 16000 - -o %d"
OGGENC (2.3 GT3b2) "-q5 - -o %d"
LAME (3.90.3) "--alt-preset standard - %d"
- All encoded with fb2k CLI and played back with same (0.8.2 on XPsp1)

But... MPC are the only affected codec of these on this particular sample.
At least I cannot hear any glitches in mp3 or mp4.



This post has been edited by Mr_Rabid_Teddybear: May 27 2004, 20:34

[Lyx]

a maybe stupid question which i asked myself multiple times, when encountering gapless problems:

If a codec uses VBR, why couldn't it just switch to maximum bitrate for the first 0.1sec and last 0.1sec of a track? I mean, pushing maximum bitrate for such a short amount of time wouldn't increase the average bitrate much.


Or wouldn't this change anything to gapless-problems?
- Lyx

[Seed]

shadowking, I don't see t2.flac attached, so I can't check this. I have the same album and there's no click between the tracks. Which program did you use to rip?

[shadowking]

http://209.152.181.168/~hydrogen/show.php/showtopic/22018

Sorry, I don't know how to attach multiple files in a message.

I used the latest EAC to rip.

[shadowking]

Specificaly EAC 0.95 pb5

[Seed]

I found t2.flac.

My test:

Decode FLACs using FLAC.exe -d. Encode using mppenc. No clicks.

Decode both FLAC files to WAV using foobar2k 0.8.2 (PCM, fixed-point) and encode both using "mppenc --xlevel --verbose t1.wav/t2.wav" (mppenc 1.15r). No clicks during playback.

Decode FLACs to WAV using "Legacy 0.7 output support : Musepack", no clicks.

Convert to MPC using foo_clienc (Commandline Encoder : MPC --standard), *click*.

One more thread where people jump to conclusions about a format without checking the problem properly first.

[shadowking]

Well I have no conclusions that's why I posted.

More tests:

Cdex to mpc 1.15r = click

Eac to Wav to mpc using fb2k 0.8.2 = click

Eac to mpc 1.14 = no click



I don't get this at all. Well its good to know that one combo is gapless.

[shadowking]

eac to Wav to mpc 1.15r using multi frontend = no click

Fb2k 0.8.2 is the strange cause ??

[shadowking]

Changing the default foobar BPS from 24 to 16 solved this sample for me...

[Pio2001]

This problem can appear with any lossy codec. Since it is lossy, the encoding of the end of track 1 can differ from the encoding of the beginning of track 2, leading to a step when the two files are played gaplessly.
The solution would be to store losslessy the first and last 50 ms of any track (50 ms being the period of a 20 Hz wave), and to perform a crossfade between the lossy version and the lossless one during these 50 ms, so that the first and last sample of all tracks are unchanged.