Timeline for Group delay compensation for non coincident drivers
Current License: CC BY-SA 4.0
19 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Mar 11 at 11:41 | history | edited | Dan Boschen | CC BY-SA 4.0 |
edited body
|
| Mar 11 at 10:48 | vote | accept | Spark123 | ||
| Mar 15, 2022 at 13:01 | comment | added | Max | You're welcome. I have a background in acoustics, so speed of sound is much closer to me, than speed of light. | |
| Mar 15, 2022 at 12:56 | comment | added | Dan Boschen | @Max Yes I see-- my thinking (and background) is with RF frequencies and speed of light...I see that even in a box we can't consider them to be collocated: 5 KHz at the speed of sound has a wavelength of only 2.6". Thanks. | |
| Mar 15, 2022 at 12:42 | comment | added | Max | Yes, a single speaker BOX. But a classical 2-way speaker has two separate speakers in it which are not collocated. | |
| Mar 15, 2022 at 12:28 | comment | added | Dan Boschen | @Max as I wrote in the last paragraph, he had clarified that the speakers are collocated and he only cares about one listener position | |
| Mar 15, 2022 at 7:37 | comment | added | Max | Keep in mind that the solution is only perfect along one single spatial plane defined by the microphone position and the normal vector of the speaker-to-speaker axis. Taking the speaker directivity into account, the plane becomes a single line (if the microphone's $z$ is halfway between the speaker's centers) or a even a single point (if it is not). The magnitude of the error created by stepping out of the plane/line/point is proportional to the distance between the speakers (in the line case) and additionally the distance microphone-speakers (in the point case). | |
| Mar 14, 2022 at 19:47 | comment | added | Dan Boschen | @Spark123 Thanks, I answered your questions directly and concisely prior to my other details. Bottom line is yes, if you consider your equalizer to correct for both phase and amplitude in the sum signal rather than just a delay equalizer (which is phase only). Be careful of noise enhancement if your dip becomes to large. | |
| Mar 14, 2022 at 19:45 | history | edited | Dan Boschen | CC BY-SA 4.0 |
added 644 characters in body
|
| Mar 14, 2022 at 16:01 | comment | added | Spark123 | I have added the plots you requested. | |
| Mar 14, 2022 at 15:35 | comment | added | Dan Boschen | Can you show the plots of each group delay as you do in the code and the result of each equalized response? I suspect you are linearizing the phase of each one, but if the result of each doesn’t have the same delay (same phase slope) you will get constructive and destructive interference in the sum. I don’t have time now to run your code line by line but if you show the plots that might make it clearer as to what is going on. | |
| Mar 14, 2022 at 15:29 | comment | added | Spark123 |
Thanks for the link, I will study it. Unfortunately I can't still understand what is wrong with the approach that led to my second question (from the code, after %% NON COINCIDENT SPEAKERS): if the group delay of the summed response is equalized to flat, why do I still have the dip in the final magnitude response introduced on purpose with the circshift? Shouldn't the group delay equalizer fix the timing issues between the woofer and the tweeter introduced by the delay and hence fix the dip?
|
|
| Mar 14, 2022 at 14:30 | history | edited | Dan Boschen | CC BY-SA 4.0 |
added 505 characters in body
|
| Mar 14, 2022 at 14:26 | comment | added | Dan Boschen | Ah hence your earlier comment on crossover. Yes are you able to characterize with a single microphone and base your equalization on that? Then you can treat your system as a black box and apply equalization according to the details in my linked post. | |
| Mar 14, 2022 at 13:49 | comment | added | Spark123 | Maybe I wasn't clear in the question, with two-way speaker I was meaning a single cabinet with a woofer and a tweeter for example, not a classical left-right stereo scenario. If I can't measure the woofer and the tweeter separately, but only together, i.e. a singole impulse response for the whole cabinet, is there a way to accomplish this? | |
| Mar 14, 2022 at 13:15 | history | undeleted | Dan Boschen | ||
| Mar 14, 2022 at 13:15 | history | edited | Dan Boschen | CC BY-SA 4.0 |
added 582 characters in body
|
| Mar 14, 2022 at 13:05 | history | deleted | Dan Boschen | via Vote | |
| Mar 14, 2022 at 12:57 | history | answered | Dan Boschen | CC BY-SA 4.0 |