Re: [Svxlink-devel] SVXLINK Improvements
Brought to you by:
sm0svx
From: Tobias B. <sm...@us...> - 2008-05-26 06:16:57
|
On Sunday 25 May 2008 23:22:01 st...@ar... wrote: > Sorry, there is a copy-and-paste bug in line 150 of ToneDetector.cpp, > please change form "lower" to "upper". That was the one I fixed. 73 de SM0SVX / Tobias > > Steve > > ----- Original Nachricht ---- > Von: Tobias Blomberg <sm...@us...> > An: Discussions about development issues > <svx...@li...> Datum: 25.05.2008 22:13 > Betreff: Re: [Svxlink-devel] SVXLINK Improvements > > > On Sunday 25 May 2008 15:14:18 st...@ar... wrote: > > > Hi Tobias, > > > > > > > Even though I would really like SvxLink to run on the NSLU2 I'm not > > > > ready > > > > > > to > > > > > > > > sacrifice the simplicity and beauty of using floats in the audio pipe > > > > concept. > > > > > > > > I have proved previously that SvxLink actually will run a SvxLink > > > > node > > > > on > > > > > > the > > > > NSLU2 at about 50% CPU usage as I remember it. I'll actually be quite > > > > happy > > > > > > > > if it only can run a remote receiver. Maybe it will manage that at > > > > 16kHz > > > > > > with > > > > floats. I don't know. > > > > > > > > The main target platform will always be a rather standard (low end) > > > > x86 PC. > > > > > > I agree that it is more complicated to do the manual scaling of audio > > > samples and it is better to invest the time required to implement it > > > for more useful things. But I think computational costs for floating > > > point emulation can be limited when the following constraints are met: > > > > > > - avoid the use of double pecision floats (type double) in the sample > > > processing code, since they are not required and its use consumes a lot > > > more computing time > > > - avoid the use of functions like division, sine, > > > cosine, exp, sqrt in the sample processing code for the same reason. > > > They should be used in the init code only. > > > > Absolutely. I just haven't gotten around to the optimizations yet and I > > like > > > > to keep it simple until I know it works and that the optimization is > > really > > > > going to make a difference. > > > > > Keeping that in mind, I implemented a new tone decoder and modified the > > > existing DTMF decoder. Also, I found out, that comparing any spectral > > > energy to the spectral energy of the entire block is not very > > > applicable, because peak detection based on the total energy depends on > > > noise and distortion at higher frequencies. The new tone decoder > > > implementation is very simple and the code self explaining, but it > > > works very relieable at less computational requirements than the old > > > decoder. The thresholds are real dB scale now, so the config parameters > > > are to be interprted slightly different. > > > > Thanks. It looks really nice and you've even complied to my coding > > standard. > > > > It gives me a warm feeling :-) > > > > I haven't checked the tone detector in detail. I'm sure it's correct. > > Just fixed a copy/paste bug in the constructor. Looking forward to > > testing it. > > > > Maybe I should rename the SpanDtmfDecoder files to something else also. > > > > I also removed dtmf_plot from compilation. It's outdated anyway. > > > > > Enclosed you willl find a patch against the existing sample_rate_16k > > > branch. > > > > Thanks. Applied and checked it in to the 16k branch. > > > > A comment about the patch, isn't it just easier to create the patch with > > "svn > > diff" instead of using the diff application with the hassle of excludes, > > mutiple source trees and so on? > > > > 73's de SM0SVX / Tobias > > > > > vy 73s > > > Steve > > > > ------------------------------------------------------------------------- > > This SF.net email is sponsored by: Microsoft > > Defy all challenges. Microsoft(R) Visual Studio 2008. > > http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ > > _______________________________________________ > > Svxlink-devel mailing list > > Svx...@li... > > https://lists.sourceforge.net/lists/listinfo/svxlink-devel > > ------------------------------------------------------------------------- > This SF.net email is sponsored by: Microsoft > Defy all challenges. Microsoft(R) Visual Studio 2008. > http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ > _______________________________________________ > Svxlink-devel mailing list > Svx...@li... > https://lists.sourceforge.net/lists/listinfo/svxlink-devel |