Rockbox mail archiveSubject: Re: Playing silence, revisited.
Re: Playing silence, revisited.
From: Michael O'Quinn <michael_at_oquinn.info>
Date: Fri, 28 Mar 2003 21:13:30 -0800 (PST)
I know, I know, it's bad taste to reply to my own post before anyone else
has a chance to. But I just realized something else that might be
important. (This also gives me a chance to correct the typo in the
I stated that the seeking speed (FF speed) doesn't affect this. That is
not entirely correct. If the magic range is reached (approx 56 seconds
into the song, I assume about where the buffer runs out) and I release the
FF key BEFORE the disk spins down, then the Silent Play Feature (tm)
But if the disk has to spin up, everything works normally, and no Silent
On Fri, 28 Mar 2003, Michael O'Quinn wrote:
> ~Hi All,
> I've been following the development of RockBox for the last month or so,
> since about 1 day after I got my Player 10 and discovered how little I
> liked the original firmware...
> Anyway, one ongoing theme seems to be "playing silence,", which (if I am
> reading this right) most folks believe has to do with faulty disk reads
> resulting in a lot of zeros being "played."
> My version is Player CVS-030326.
> I was playing a 256 CBR song, and encountered the Silent Play Feature
> (tm), so I decided to see if I could reproduce it and observe anything
> that might be useful.
> It turns out I can reproduce it on demand, every time with any song on
> that album. (They are all 256 CBR.) Here's how. I jump to the beginning
> of the song, and then while this disk light is still on (actually might
> work if I wait a bit, didn't test that) if I immediately FF to between 0:52
> and 0:58 in the song, the red light comes on, and it plays silence, but
> the seconds keep counting up.
> After a random time, between 1:22 and up to the next buffer refill, the
> music starts again, but usually WITHOUT reading the disk. No spin up
> vibration, no red light, the music just restarts by itself. A couple of
> times it went until somewhere around 2 minutes into the song, and the
> music restarted after a disk spin up and read.
> Time to FF doesn't seem to affect it. At first my FF was set at min step
> = 2 sec, and accel = 2x/1 sec. I eventually ramped that down as slow as
> possible (min step = 1 sec, accel = off) to get more accurate, and it
> didn't seem to affect it. (Except, of course, that the behavior got more
> consistent as my seeks got more accurate...)
> One thing that affected this is that most times when I would jump to the
> beginning of a song, it would start counting up and playing from 0:00.
> Sometimes, however, there would be a pause, and it would start counting
> and playing from between 2-5 seconds into the song. (This is actually a
> minor bug in itself -- beginning of the song shouldn't be chopped off like
> The Silent Play Feature (tm) would only activate when the song started
> playing from 0:00, and NEVER when it started a few seconds into the song.
> What seems significant about this is that, in terms of reads and filling
> buffers, it seems that the BEGINNING of the buffer that's getting munged,
> not the END.
> This isn't a cry for help. I'm very satisfied (thrilled actually!!!) with
> RockBox, to the point that I forget the actual device is a "Jukebox", not
> a "RockBox". Rather, I've been looking for a way to get involved and
> help, and I figure a decent bug report is one way to get started. I could
> provide one of the songs, or do more testing if someone wants..
> I hope this helps in tracking this one down.
> Michael O'Quinn
Received on 2003-03-29