On 2009/03/19 04:23, a...@alanlee.org wrote: > > Hi, long time lurker, first time poster.. I just wrote a first cut of a MPD > output driver for Sigma Designs smp86xx SoCs - specifically Networked Media > Tank appliances. While testing today I noticed a few things that seemed like > bugs and just wanted to confirm if things are working as intended. > > 1) When providing 24-bit sample output - mpd really doesn't. It > provides 24-bit amplitutes that are signed extended out in 32-bit > samples. Is there a way to turn 24-bit sample packing on?
No. MPD has gained 24 bit support only recently, and very recently also 32bit support (in 0.15~git). If you need packed 24 bit samples, convert them manually (the pcm_buffer library may be helpful). We will eliminate your code as soon as MPD's core supports that format. > 2) While I don't think it's a bug with MPD, I was curious if anyone > else has experienced issues with libmad compiled with the accuracy > switch producing horrible overdriving and clipping pops when the > output is forced to 16-bit samples? It seems fine when I repack the > 24-bit output but not when I override the output to 16-bit. This is > on a MIPS 4000 w/ libmad .15.1b and a mpd update from git current as > of today. No, havn't noticed that yet, but you might try to run the standalone test tools in ./test/ - e.g. ./test/run_decoder to dump the raw libmad output, and maybe write a ./test/convert program to convert that raw 24 bit output to 16 bit - then analyze the output file for problems. Might be a bug in MPD's conversion code. > 3) I appears to me the only thing putting throttle back-pressure on > the MPD upper framework is the size of the drivers output sample > buffer and any waiting it has to do for available buffer space. This > really really seems lacking. Especially when it comes to playing > streaming sources and differences in crystals from the stream and > playback device. Was just curious if this is just where the design > is at atm (and I know I should feel free to improve it!) Some more > (than a PC) advanced players often have slewable clocks and vcxos to > deal with that problem. Though Internet jitter may make clock > recovery moot. You are right, this is a shortcoming, and is something we need to improve. We havn't made up an idea yet, and suggestions are welcome. Max ------------------------------------------------------------------------------ Apps built with the Adobe(R) Flex(R) framework and Flex Builder(TM) are powering Web 2.0 with engaging, cross-platform capabilities. Quickly and easily build your RIAs with Flex Builder, the Eclipse(TM)based development software that enables intelligent coding and step-through debugging. Download the free 60 day trial. http://p.sf.net/sfu/www-adobe-com _______________________________________________ Musicpd-dev-team mailing list Musicpd-dev-team@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/musicpd-dev-team
