1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.

    Dismiss Notice

Recording Impulse Responses for Speaker Virtualization

Discussion in 'Sound Science' started by jaakkopasanen, Oct 9, 2018.
First
 
Back
1 2 3 4 5 6 7
9 10 11 12 13
Next
 
Last
  1. jaakkopasanen
  2. Dasm
  3. arksergo
    No, I had not. I get very good result with first try.
     
    Last edited: Nov 5, 2019
  4. jaakkopasanen
    There are no info about frequency response, sensitivity or noise.

    Have you made more than one measurement. I'm wondering how reproducible your results are. I have one good stereo measurement with close to perfect stereo balance but it's really hard to reproduce this success with my mics.

    I might cut open the hooks of my mics to take the capsule out and place it in an ear plug. That would be the fastest way for me to test my hypothesis.
     
  5. Dasm
    Freq responce is easy to compensate. I bought 5 pcs, will show you measurements later.
     
  6. arksergo
    No, I measured once. I am going to do some additional measurements this weekend, wanna try to get big virtual room for movies and separate layout for 5.1 movies
     
  7. 71 dB
    I tried electret mics Jecklin disc and ear mics, but it's not working. Signal to noise ratio is complete rubbish. Weird noises like ghost hunter gear! Useless! My first try with mics, so I probably made all the mistakes. Depressing.

    Circuit-using-Electret-Condenser-Microphone.png

    Using this circuit and feeding Olympus LP-5 recorder ( I have tried both line in and mic connection). I don't know what is wrong. Would this help:

    images.jpg
     
    Last edited: Nov 5, 2019
  8. johnn29
    Is there a way we can ask Impulcifer to assume channel balance with the headphones and then attenuate L or R to get balance? When I correct for it via HeSuVi's attenuation I'm hard pressed to find anything "wrong" with the sound. It sounds really good and is bang on center then. Again I A vs B against the real speakers and they're remarkably similar. It's not identical though, but even my naturally perfect measurement wasn't identical.

    Only thing I find frustrating playing with the balance is then I can't A vs B with my perfect recording quickly. Although if I can't pick a winner with a 5 second pause, perhaps that means there's no real better version!
     
  9. arksergo
    I had hard times trying to get good signal from my mics too. In fact it was the most challenging part for me, because I am not any good in analog schematics.
    Check out this link: http://www.epanorama.net/circuits/microphone_powering.html
    Also you can find good scheme for measurement mic here (russian only): http://cxo.lv/index.php/solder/micamp/111-micamp01, author suggests to use this scheme:
    [​IMG]

    I used this scheme, modified for balance output and phantom power with lm358 op amp. It is quite good, but you should keep wires from mics to amp as short as possible. Mine was about 20 cm and I ended up using it like this:
    [​IMG]
     
    castleofargh and 71 dB like this.
  10. jaakkopasanen
    That shouldn't be too much work. Would you care to create an issue about this?
     
  11. jaakkopasanen
    Small update: headphone plots are now produced befor HRIR processing. This makes it possible to run Impulcifer right after headphone compensation recording to check the left and right ear frequency responses. This will make it faster to iterate with the mic placement because now headphone recording - processing - mic adjustment cycle can be done before even the first HRIR recording has been made. So in this stage the folder only needs to contain headphones.wav and the plots folder will have Headphones.png.

    This allowed me to do a new measurement with good channel balance.
     
  12. 71 dB
    Thanks! Your gear looks badass! :)
     
  13. Joe Bloggs Contributor
    You can always assume that the left and right ear should receive HRIRs with the same frequency response and hence, for any measurement where there's a left and right ear version,
    1. Calculate average of responses from the two ears
    2. Equalize the impulses for the left and right ears to the average response, using minimum phase filters. The EQs should be octave smoothed to a certain degree so that the filters don't "try too hard".

    This will yield a very balanced soundstage in most cases yet sound much superior to if you'd simply made the left and right impulses exactly the same. I know, I've been doing it this way for years.

    Incredible idea, isn't it :ksc75smile:
     
    Last edited: Nov 9, 2019
  14. jaakkopasanen
    I've been thinking about this but wasn't sure if that would be ideal since not everyone's ears are symmetrical. Maybe this is not ideal but good enough. At least I could add a parameter for doing this so users could try it out easily and hear if it improves the result in their case.
     
    Joe Bloggs likes this.
  15. jaakkopasanen
    Busy day. Got plenty of tools for solving channel balance issues.

    Firstly the headphone and result plots have now both sides in the same graph for easier comparison. This will make it easier to play around with mic placement and to figure out what is wrong with the channel balance.

    The bigger change is channel balancer tool which adjusts left and right side levels or frequency responses to match each other. This usually nails the channel balance but terribly bad recordings cannot be made perfect. Here is the documentation about the new feature: https://github.com/jaakkopasanen/Impulcifer#channel-balance-correction

    See here for squiggly lines: https://imgur.com/a/cWfO8Zt

    I also made mock microphones by gluing a 11mm disc to a foam ear plug to see if 10mm or so microphone would fit into the ear canal opening. I doesn't and I have fairly large ears. I figured out that FEL Communications sell also pre-made EM258 mono modules and RODE VXLR+ adapters. Two of both would be needed which would make about 100€.
    https://micbooster.com/primo-microphone-capsules/97-primo-em258-mono-module-with-35mm-plug-10m.html
    https://micbooster.com/leads-and-ad...r-converter.html?search_query=vxlr+&results=1
    I sent them an email asking if they could provide binaural mics with matched EM258 modules and a ear plug style support and possibly adapters to make them work with both recorders and USB audio interfaces.
     
    Joe Bloggs likes this.
First
 
Back
1 2 3 4 5 6 7
9 10 11 12 13
Next
 
Last

Share This Page