#asahi · 2026-09-04 13:32
waltmckfrom what I've read you really should be bucketing frequencies if you do mvdr#asahi · 2026-09-04 13:32
waltmckstill definitely not ideal though#asahi · 2026-09-04 13:32
waltmckI also wrote a robust variant of chadmed's beamformer about a year ago with some performance optimizations: https://github.com/waltmck/triforce#asahi · 2026-09-04 13:31
ashafqI should've hold on to that for beamforming work.#asahi · 2026-09-04 13:31
chadmedanything M1-M3 will work :) #asahi · 2026-09-04 13:31
ashafqchadmed: now I regret swapping out my wife's M1 macbook :)#asahi · 2026-09-04 13:30
chadmednormal rules apply, no slop and i would like this to stand as a freely available gold-ish standard so please make sure everything you do is well documented and purposeful#asahi · 2026-09-04 13:29
sven:D#asahi · 2026-09-04 13:29
ashafqchadmed: I use my closet to do these measurements.#asahi · 2026-09-04 13:28
chadmedfeel free to hack on it and do whatever #asahi · 2026-09-04 13:28
Tachyonray[m]there are good open metrics to evaluate the quality of the voice band#asahi · 2026-09-04 13:28
chaos_princesswhich device(s) do you have#asahi · 2026-09-04 13:28
ashafqa hole acts like a low pass filter, so you have to account for it in beamformer design.#asahi · 2026-09-04 13:28
chadmedand re speed of sound i did everything at standard room temp conditions (as well as i could control, i dont have a real lab) #asahi · 2026-09-04 13:27
chadmedit works acceptably for voice band which is all we really care about here#asahi · 2026-09-04 13:27
chaos_princessLike, it is not some oem sludge that bows when you pick it up by a corner#asahi · 2026-09-04 13:27
ashafqchaos_princess: I don't know. I will have to measure raw recordings to find out.#asahi · 2026-09-04 13:27
Tachyonray[m]also small changes in the speed of sound can be significant#asahi · 2026-09-04 13:27
chaos_princessisn't the chassis on macs stiff enough so that it should _mostly_ not matter?#asahi · 2026-09-04 13:26
chadmedyes like i said, very naive implementation and the geometry was basically done by eyeballing and scaling pictures of replacement parts#asahi · 2026-09-04 13:26
ashafqthe ports/plastic/material can alter length of the array, so you have to take that into account.#asahi · 2026-09-04 13:26
chadmedand that is to say nothing of the insane freaks working in RF land #asahi · 2026-09-04 13:25
chadmedthere are heaps of toy implementations that people put out as part of uni projects but everything at the leading edge is secret and obfuscated #asahi · 2026-09-04 13:25
ashafqit's also black magic :)#asahi · 2026-09-04 13:25
chadmedi hate that there is basically no publicly accessible literature on any of this stuff because its all either commercially sensitive or state secrets #asahi · 2026-09-04 13:24
Tachyonray[m]that’s awesome, i did my undergrad thesis on acoustic mvdr beamformers#asahi · 2026-09-04 13:24
chadmedits a triangular array in every supported mac#asahi · 2026-09-04 13:24
ashafqtypically you have more than one beamforner to get the "best SNR signal"#asahi · 2026-09-04 13:24
chadmedthe array geometry is described in the asahi-audio repo#asahi · 2026-09-04 13:24
ashafqI have to figure out where the microphones are#asahi · 2026-09-04 13:23
chadmedyep #asahi · 2026-09-04 13:23
ashafqchadmed: this is microphone beamformer?#asahi · 2026-09-04 13:23
chadmedit works kinda but its not very good #asahi · 2026-09-04 13:23
chadmedit is a very naive attempt at an MVDR beamformer using first year undergrad engineering dropout knowledge#asahi · 2026-09-04 13:22
chadmedyou may want to take a look at https://github.com/chadmed/triforce#asahi · 2026-09-04 13:22
chadmedashafq: sorry i meant to say something earlier today but got swamped #asahi · 2026-09-04 13:22
chaos_princessa lot is already working, but not like it can't be improved#asahi · 2026-09-04 13:21
chaos_princessprobably get in touch with chadmed#asahi · 2026-09-04 13:19
ashafqwaltmck: yes, I have 10+ years of audio dsp experience.#asahi · 2026-09-03 22:38
arcathraxhi#asahi · 2026-09-03 20:52
waltmckashafq: do you have audio dsp experience?#asahi · 2026-09-02 17:50
MauricioCuevas[m]oh, its' okay, thank u, I'll check if it's possible to reduce the time#asahi · 2026-09-02 17:47
fl0_id_MauricioCuevas[m] no. in case there is a linux problem with grub or sth I think#asahi · 2026-09-02 17:47
fl0_id_usb-c display will probably be more broadly available at some point, but there is no date#asahi · 2026-09-02 17:47
MauricioCuevas[m]to get access to macOS? #asahi · 2026-09-02 17:46
fl0_id_MauricioCuevas[m] I think you could just make the display time very short. but in some cases you might need it#asahi · 2026-09-02 17:41
MauricioCuevas[m]Also is there a way to not get the U boot screen when im booting? the one with the submarine lol#asahi · 2026-09-02 17:40
MauricioCuevas[m]aw :( #asahi · 2026-09-02 17:39
fl0_id_MauricioCuevas[m] it's not officially supported. there is an experimental development kernel, but you would be on your own to install or debug it if it does not work#asahi · 2026-09-02 17:32
MauricioCuevas[m]yes! is it possible?