From: Steve Harris (S.W.Harris_at_ecs.soton.ac.uk)
Date: 01/01/03 06:39 EST
From: Steve Harris <S.W.Harris@ecs.soton.ac.uk> Subject: Re: [ardour-dev] waveform display query Message-ID: <20030101113919.GA19179@ecs.soton.ac.uk> Date: Wed, 1 Jan 2003 11:39:19 +0000 On Tue, Dec 31, 2002 at 09:35:40 -0500, Paul Davis wrote: > > > more to the point, you can *add* those tools to ardour... > > > >Indeed! I've learned my way around gtk+ while working on gEDA, so I > >guess I'll implement this feature if I'm not stepping on anyone's > >toes. (The other thing I intended to work on was a shaped-noise > >dither plugin, unless someone else has started on one.) > > you mean like the one steve harris wrote for JACK's ALSA client/driver? I think this should probably be turned on by default for 16bit devices, maybe just triangular, but some sort of dither. > he used triangular waveshaped dithering for export, because i asked > him for something cheaper than shaped noise, IIRC: This code is wrong. To fix it it requires some sort of per channel state for the export code (like in the JACK code). Changing tmp += (random() & 0x3F) + (random() & 0x3F); to tmp += random() & 0xFF; will improve it, by changing it to rectangular, but its not a real fix. Looking at it, I think the right thing to do is to always generate the float output and then generate the dithered, 16bit out from that if its needed. I dont know why I didn't do that before. does the dataF[] contain the final output? If so I will fix this code this afternoon. By the time its gets to ExportDataReady_(short|float) is all the output accumulated? > if someone wants to make a case for shaped noise dithering in export, > please let me hear it. it would be easy to replace what we have now. It should probably be an option, for some cases it would be superiour, it should be easy to use the code in JACK. > a LADSPA dither plugin might be useful for some situations, but its > not clear how useful it is for Ardour ... Its not realy meaningful, dither needs to be float->int. Baring hacks like [-1,1] in, [-32768,32767] out I dont see how you could guarantee that the float->int part of the host would pick the right bits. - Steve ------------------------------------------------------- This sf.net email is sponsored by:ThinkGeek Welcome to geek heaven. http://thinkgeek.com/sf _______________________________________________ ardour-dev mailing list ardour-dev@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/ardour-dev
This archive was generated by hypermail 2.1.5 : 01/01/03 06:47 EST