Fade on stop/pause - via Line-out

[brenthaag]

I have the same problem/request.  I would like to point out that when you skip tracks (and the fade-in/fade-out is turned on), it will fade out the playing song and fade in the new song on the line out--at least it seems to on my iPod color.

[karashata]

I would like to point out that when you skip tracks (and the fade-in/fade-out is turned on), it will fade out the playing song and fade in the new song on the line out--at least it seems to on my iPod color.

You may want to check if you have crossfade on (set to either "Track Skip Only" or "Shuffle and Track Skip"), since that's what crossfade is supposed to do...

[Falco98]

I would like to point out that when you skip tracks (and the fade-in/fade-out is turned on), it will fade out the playing song and fade in the new song on the line out--at least it seems to on my iPod color.

You may want to check if you have crossfade on (set to either "Track Skip Only" or "Shuffle and Track Skip"), since that's what crossfade is supposed to do...

I'm pretty sure that's exactly the point he was making.  In other words, there are no problems with fading the audio out for a crossfade, but it won't fade out when pausing.  Isn't that a rather simple principle?

I agree with this issue - I just got my 5.5g 80gig ipod and noticed this problem pretty much immediately.  Also I was wondering whether the fade-to-pause is really supposed to take what seems like 2 seconds, since on my iRiver 140 it was always nearly instantaneous.  But I digress:  why can't the fade-on-pause be handled in the same way as the fade out for crossfade (which, once again, works)?  I'm not familiar with the terminology, but it seems some others here were calling this "dsp". 

Also, it seems like it would be nice to be able to set the fade length in milliseconds, or something; i want a fade on pause to avoid a harsh cut-off, but I don't need a 2-second fadeout.

[JdGordon]

fade on pause/stop is currently hard coded and implementing using the output volume. crossfade is implemented using the DSP (digital signal proccessing... software which mangles the output).
If someone were to redo fade to use the DSP then it would work on the lineout and be the same as crossfade. noone has stepped up to do that.

[saratoga]

Using DSP for L/R balance would also be nice so that balance can be adjusted over line out.

[JdGordon]

so someone just needs to come up with a really fast DSP so we can do this stuff at the last possible moment before it goes out to the DAC

[Multiplex]

so someone just needs to come up with a really fast DSP so we can do this stuff at the last possible moment before it goes out to the DAC

I'm assuming that comment is for the fade on pause - L/R balance would be OK using the same mechanism as the other interactive DSP operations - or even the ones that have no special treatment (Replay Gain Pre-Amp for example)

I remember reading some of the code that deals with that and being HUGELY impressed - I've worked on 'professional' projects where the quality of the code was nowhere near as good as Rockbox!

[Blue Dude]

The problem with moving instantaneous user interface effects such as fade-on-pause, key clicks, volume changes, balance, etc. to the DSP is that the DSP is not at all instantaneous.  Depending on the size and mode of the PCM buffer, it can lag by several seconds.  Even setting low latency means an unacceptable 1/4 second or more delay.

Crossfade sounds smooth because it is uses a less granular fade interval than the volume control, but its fade effect is mixed directly into the output buffer, usually with a several second delay.  Quite a lot of trouble goes into making a fade happen in real time.  It's possible to do a real time pause fade in the same fashion, but it would be destructive, requiring a rebuffer every time it's done (effectively equivalent to stop/resume).  This isn't prohibitive, but it's one reason why it's not a trivial change.  It's just plain easier to step the volume down, pause the DMA in hardware, then step up than it is to operate directly on the output buffer, stop everything, and then resume.

Sounds like an interesting project, though...

[Falco98]

I have checked the changelog (major changes, anyway) and there is no mention of this, but the issue seems to have been fixed in 3.5 -- i am running a 5.5g 80gb ipod, and now when playing thru its line-out in my girlfriend's car, it does fade out and in on stop and pause / resume.  I wish it were mentioned in a changelog somewhere, I'm curious as to what was done.