Crossfading - what can I do about that?

[Greg Trounson]

On 31/08/10 05:19, Jeff Mitchell wrote:
> On 8/30/2010 12:58 PM, Nikos Chantziaras wrote:
>> On 08/30/2010 07:49 PM, Jeff Mitchell wrote:
>>> On 8/30/2010 8:10 AM, Thomas Lübking wrote:
>>>> Am Monday 30 August 2010 schrieb Nikos Chantziaras:
>>>>> It does *not* work with Ogg Vorbis.  It never did.  It doesn't work with
>>>>> *any* format.
>>>
>>> Right, so, one thing to keep in mind is that it's always been
>>> engine-dependent. Amarok supports it if the engine supports it.
>>> Different engines may have different bugs/features regarding gapless --
>>> that was true in the 1.4 days and still true with Phonon AFAIK.
>>
>> Doesn't work with Xine and neither with VCL backends :-/  Last time I
>> tried GStreamer, I remember it worked fine there.
>>
>> The Xine devs assured everyone asking about it that Xine supports
>> gapless 100%.
>
> Entirely possible the weak link is Phonon, then.
>
>>>> have no relevant Vorbis, but (just tested) it does work with FLAC - and
>>>> actually it does (in a hackish way) work with even mp3, however afaics
>>>> (either) _only_ with the xine backend :-( - gstreamer stopped working at all -
>>>> and i don't care :-) and there's a minor gap with vlc and a bigger one with
>>>> the mplayer backend ->   bug phonon
>>>
>>> You're right about MP3/Xine...Xine actually hacked together fake pseudo
>>> gapless support for MP3 but from what I know it was always rather hacky.
>>
>> MP3s encoded with LAME as opposed to other encoders are supposed to be
>> perfectly able to have gapless support; not sure if Xine implements this
>> LAME standard though.  I don't have many MP3s to test with.  The vast
>> majority (>90%) of my music is Vorbis.
>
> Xine does support it. Also keep in mind that "perfectly able" means "it
> mostly works" but it's still not true gapless.  :-)

I've done some tests on a Fedora 13 x86_64 box with a single music track 
split into two WAV files, then two OGG files.  Here's how each 
application handles the transition from file_1 to file_2.  In no 
particular order:
(gap times are my best guess)

Xine 0.99.6 WAV: 	gapless
Xine 0.99.6 OGG: 	0.2s gap of silence
mplayer SVN-r31628-4.4.4 WAV:	0.5s gap of silence
mplayer SVN-r31628-4.4.4 OGG:	0.5s gap of silence
amarok 2.3.1 WAV:	gapless
amarok 2.3.1 OGG:	0.1s gap of LOUD noise
ogg123 1.2.0 WAV:	not supported
ogg123 1.2.0 OGG:	gapless
VLC 1.0.6 WAV:		0.8s gap of silence
VLC 1.0.6 OGG		0.8s gap of silence
audacious 2.2-16 WAV	0.2s gap of silence
audacious 2.2-16 OGG	0.2s gap of silence
rhythmbox 0.12.8 WAV	gapless
rhythmbox 0.12.8 OGG	gapless

This was all done with the default output (usually pulseaudio). 
Audacious offered a gapless ALSA output plugin, but this tended to 
overlay static on the music.  Testing was carried out first on an 
NFS-mounted volume, then a local EXT4 volume with identical results.

Amarok did very well with the WAV and had a shorter OGG playback gap 
than others, but the gap consisted of a segment much louder than the 
song itself, about 0.1s of what could be the previous 0.1s repeated.

Greg