#asahi-dev · 2022-03-14 09:33
chadmedif you can reliably do all that in software, you can use a very basic paper/cardboard cone, ditch the LC/RLC filters#asahi-dev · 2022-03-14 09:32
chadmedwell its a bit of both. modern loudspeakers are usually made of pretty exotic materials and have some decently advanced passive (sometimes even active) electronics in them to control things like crossover frequencies and such #asahi-dev · 2022-03-14 09:29
chadmedwe're not, ive added them as a chain of pipewire convolutions#asahi-dev · 2022-03-14 09:29
chadmedso its been interesting seeing the other side of it, where its used as basically a cost optimisation strategy#asahi-dev · 2022-03-14 09:28
chadmedmy background in terms of audio stuff is coming from professional recording gear and hifi stuff where the engineering is in the physical hardware, and your goal is to apply as little DSP as is absolutely neccessary to correct for the room's characteristics (if at all)#asahi-dev · 2022-03-14 09:27
chadmedoh yeah theres nothing wrong with it #asahi-dev · 2022-03-14 09:25
chadmednot because its a bad thing or whatever, just because i thought it was always super impressive that they could "make" such small speakers that put out a decent sound. turns out they cant! :P #asahi-dev · 2022-03-14 09:24
chadmedi was devastated when i first learnt that like 90% of what makes tiny speakers like this sound good is the DSP in the background #asahi-dev · 2022-03-14 09:23
chadmedwe will have to if we want them to sound good#asahi-dev · 2022-03-14 09:22
chadmedbut they wont sound very good#asahi-dev · 2022-03-14 09:22
chadmedwell, you could apply them and see i guess#asahi-dev · 2022-03-14 09:21
chadmedyeah my transforms wont work though#asahi-dev · 2022-03-14 08:55
chadmednow that i have the workflow down and have wrapped my head around playing with alsa and pipewire, i could probably manage a turnaround of t+1d from the kernel driver being made available if i had access to the machines#asahi-dev · 2022-03-14 08:49
chadmedi mean eventually it will work on the j313, eventually it will work on all the machines #asahi-dev · 2022-03-14 08:46
chadmedjoske: damn i should change my naming scheme then heh#asahi-dev · 2022-03-14 05:37
chadmedim pretty happy with this for a release tbh, only thing im still not happy with is that the bass completely disappears at low volumes, but we'd need some sort of dynamic loudness/expander filter to do that and i've no faith that pipewire could do that properly#asahi-dev · 2022-03-14 05:36
chadmedpovik, marcan, anyone else brave enough to test their j31x speakers: https://drive.google.com/file/d/1r--1iLggAO2JII-5UFJYw7_fTrD5dvKl/view?usp=sharing#asahi-dev · 2022-03-14 04:42
chadmedwhich actually works#asahi-dev · 2022-03-14 04:42
chadmedand the worst part is it turns out that wireplumber was the problem causing the filter chain to not route automatically to the output, so ive had to fall back on good ol pipewire-media-session#asahi-dev · 2022-03-14 04:41
chadmedyeah its a real shame#asahi-dev · 2022-03-14 04:09
chadmedwhich is obviously not what we want to do here#asahi-dev · 2022-03-14 04:08
chadmeddamn so you cant actually hide devices with pipewire/wireplumber, only disable them completely#asahi-dev · 2022-03-13 07:08
chadmedi would very much like it if we could get FIR storage standardised somewhere like /usr/lib/alsa/impulse/VENDOR/BOARD/ and shipped as part of alsa-lib but thats above my head#asahi-dev · 2022-03-13 07:06
chadmedhopefully all we need to ship is a custom pipewire config fragment#asahi-dev · 2022-03-13 07:05
chadmedi just need to work out how to get it to route all sound via the filter chain by default and hide the physical device from clients so they never connect to it #asahi-dev · 2022-03-13 07:05
chadmedwe're in business for pipewire#asahi-dev · 2022-03-13 04:52
chadmedi screwed up the crossovers#asahi-dev · 2022-03-13 04:52
chadmednah it was a PEBKAC issue :P#asahi-dev · 2022-03-13 04:06
chadmedhuh nvm it wasnt the phasing... bass seems to come and go with reboots? #asahi-dev · 2022-03-13 03:16
chadmedif i invert the phase of the "sub" woofers and attenuate them a little bit, all the bass comes back #asahi-dev · 2022-03-13 03:15
chadmedthe woofers are out of phase with each other#asahi-dev · 2022-03-13 03:15
chadmedoh and i figured out why the bass response was so dead with all the drivers enabled#asahi-dev · 2022-03-13 02:33
chadmedusing the filter chain in pipewire just makes the whole thing die, and using them in alsaimpulse results in silence on whatever channels theyre applied to #asahi-dev · 2022-03-13 02:33
chadmedbut pipewire and alsaimpulse just will not accept them#asahi-dev · 2022-03-13 02:33
chadmedso i trimmed the impulses i got from REW to remove silence and noise, and they work just fine in carla, make a huge difference#asahi-dev · 2022-03-12 12:12
chadmedapplying just two of the IRs makes a significant difference though, so we've made progress at least#asahi-dev · 2022-03-12 12:11
chadmedi'll do that tomorrow though, now that i have a flow that works its quick and easy to do stuff#asahi-dev · 2022-03-12 12:10
chadmedok change of plans for the IRs -- im going to sample just L and R, with the crossovers applied and use those samples to build an IR. trying to apply 6 at once explodes ALSA and pegs a whole core#asahi-dev · 2022-03-12 08:51
chadmedof course you could apply an EQ curve to a file or something, but if the output of your filters is going to speakers, you can only really do it as the samples arrive#asahi-dev · 2022-03-12 08:50
chadmedwell i mean, if youre driving speakers live that is#asahi-dev · 2022-03-12 08:50
chadmedaudio signals are time-dependent, theres no such thing as non-real-time DSP #asahi-dev · 2022-03-12 08:49
chadmedits the only way to do it lol#asahi-dev · 2022-03-12 08:49
chadmedthat said i dont really have the money or time to do this for each and every new machine that comes out, as much as id love to :P#asahi-dev · 2022-03-12 08:48
chadmedthe hardest part is forging the path #asahi-dev · 2022-03-12 08:48
chadmedim just using the j314s as a proof of concept and to establish a process going forward for new machines#asahi-dev · 2022-03-12 08:48
chadmedyeah this is going to have to be a per-machine thing#asahi-dev · 2022-03-12 08:45
chadmedhopefully some magic#asahi-dev · 2022-03-12 08:45
chadmedim hoping so. i want to see what happens when i apply them to the drivers as is using brutefir or something similar#asahi-dev · 2022-03-12 08:30
chadmedhttps://drive.google.com/file/d/122M6wB8eRCw6A1FM85i8edd9AyCNCxTQ/view?usp=sharing <- 24/96 WAV impulses for each driver#asahi-dev · 2022-03-12 07:18
chadmedsounds to me like theyre expanding the mids. one thing it does better though is the tightness of the bass response. its pretty dead under linux regardless of what i do with the EQ, which makes me think theyve got some funky phasing going on to overcome standing waves set up in the chassis