On Tue, Jan 08, 2008 at 09:08:46PM +0100, Adam Sulmicki wrote: > Hello, > > I was wondering. Is there some way to do the following in PulseAudio: > > SELECTIVE SOUND AMPLIFICATION BASED ON SOUND FREQUENCY. Yes, PulseAudio can do equalization (which is the established term for this), with help from some appropriate LADSPA plugin. Documentation for loading LADSPA plugins to the pulse server by editing default.pa (located in /etc/pulse/ or ~/.pulse/) is available here: http://www.pulseaudio.org/wiki/Modules#module-ladspa-sink If you aren't familiar with doing manual pulse configuration, feel free to ask further questions. Also, as pointed out in the module documentation, experimenting with different parameter values isn't necessarily exactly easy. If all you want to do is to do amplification as shown in the link you provided, it's easier, just use the same plugin as in the example (mbeq) and assign appropriate amplification values to the control values. The 15 numbers mean amplifications in decibels, and map to the following frequencies: 50 Hz and less 100 Hz 156 Hz 220 Hz 311 Hz 440 Hz 622 Hz 880 Hz 1250 Hz 1750 Hz 2500 Hz 3500 Hz 5000 Hz 10000 Hz 20000 Hz There's one more thing adding complexity. Digital amplification may cause distortion to the signal, if the input is already maximally loud (which is the usual situation). Therefore I recommend adding this line to default.pa: load-module module-match and then create file ~/.pulse/match.table with the following line: .* 32000 As can be read from the same Modules page I already linked, module-match sets playback streams' volumes automatically. This particular line in match.table causes the volume to be set to about 50% (the volume value range is 0-65535). 40dB is quite a lot of amplifying, so if it still causes distortion, lower this value further. You mentioned that overall 10dB amplification on top of the frequency-specific amplification would be desirable (like the diagram showed). That can of course be accomplished by turning the knob on the speakers :) Assuming that they have one, that is... > Can this be done on PC under Linux? without using DSP? I tried to Google for > this but I got no useful hits. The example I linked is from my own default.pa, and I use it all the time. This is a PC machine and I'm running Linux, so yes, it can be done. By DSP I assume you mean a separate DSP card (or some ultra fancy sound card) - a DSP card isn't required, equalizing doesn't eat CPU power awfully much. Hope this helps. -- Tanu Kaskinen
