SBAF GitHub: Headphone Frequency Response Data

Discussion in 'Audio Science' started by purr1n, Aug 2, 2020.

  1. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,777
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    In light of the contributions, I will be making all my headphone data available on public GitHub:

    HPDB - The Headphone Database project

    Http://hpdb.io

    - or -

    https://github.com/superbestaudiofriends/headphone-measurements

    upload_2020-8-2_12-31-38.png

    upload_2020-8-2_12-23-52.png

    I've thrown up a few files, so if you guys can test them, that would be great. I put up a bunch of files earlier only to realize that a few numbers toward the top of the frequency range got garbled (I wrote some code to normalize the frequency response files for their intended purposes). Note that with some headphones, I only have the result from one channel.

    One step at a time, so I want to start with frequency response first. This will be the most useful for people who want to create their own EQ profiles. The headphone frequency data I have put up there is compensated for a perceptual neutral as a straight vertical line across. For now, this data is copyrighted. However, can any of you guys who could tell me exactly what I need to do in concise steps to make it Creative Commons with Attribution? I don't care how the data is used, even if people want to make money from it with fancy EQ plug-ins. It would be great if someone can take the data and make a frequency response comparator. I'm too busy and most of my coding has been for tougher things involving numbers, e.g. attack and decay envelope for burst responses, etc. I know some of you guys have mentioned how you can help with coding stuff. Here is your opportunity.

    All measurement data will be compensated, but I will leave links with the reverse curves so that one can undo the compensation. I'm sure many will want to develop their own. Long term I plan on getting my legacy data from the flat plate coupler and "adapt" them to an SBAF comp. In some instances, I will provide both flat plate and miniDSP ears. For the sake of consistency, I will only provide compensated FR plots in a way which I feel best approximates perceptive neutral. Another reason for this is because the Harmon target is getting way too much traction by people who are smart but lack experience (Rtings, Oratory, etc). It's possible that the Harmon curve is indeed a curve that reflect consumer tastes, particularly with IEM uses in Asia*. However, I want a target that is similar to what sound engineers are accustomed to in the mixing and mastering studios near where I live in the LA basin.

    This will be a long term project, so feel free to suggest any ideas. Finally, any ideas on a brand name for this project?

    UPDATE:

    Use this tool to visualize the data: http://billp.site/projects/chart-tool/

    Because 500 to 1500Hz is normalized to 0db in the data, yype 90 for the offset and press the "+" button next to the offset field so it displays properly.

    * Next time I hear someone say Harmon "neutral" target, I'm gonna rip someone's balls off.
     
    Last edited: Aug 21, 2020
  2. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,777
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    If there is anything "missing" that you would like see to be uploaded up there, please make a request.
     
  3. Biodegraded

    Biodegraded Friend

    Pyrate Contributor
    Joined:
    May 28, 2017
    Likes Received:
    7,982
    Trophy Points:
    113
    Location:
    Vancouver BC
    Great! That format comes into REW just fine using File / Import / Import frequency response. Maybe the zero dB frequency and the smoothing should be included in headers - or if smoothings all are going to be 1/6, mention that in the Readme file.

    One hassle in the current REW release is that it's not possible to apply a compensation/calibration to imported data, or files saved from imported data. Can't remember if I read that JohnM had added that to one of the betas, but it's easy enough to do in Excel before import if people want to.

    Edit: No Drop Panda R?

    Edit 2: Was going to give you grief about the Elex L data but I see you've removed it from the current set :D
     
    Last edited: Aug 2, 2020
  4. mawk

    mawk Acquaintance

    Contributor
    Joined:
    May 23, 2016
    Likes Received:
    89
    Trophy Points:
    18
    Regarding licensing: All you really have to do is state in the readme that the work is licensed under "Creative Commons - Attribution 4.0 International", and optionally provide a link to the license terms (https://creativecommons.org/licenses/by/4.0/legalcode). It's up to everyone else to make sure they've read the terms, you have no other obligations.
     
  5. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,777
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    Yup, details I forgot. All files will be 1/6 octave smoothed - noted this in the README.md

    I do wonder if tab delimited instead of space delimited may be better because the cut and paste into sheets automagically goes into two separate columns this way. Let me know if tab is preferable ASAP.

    For some headphones and calibrations, I only did or could only find one channel of data. It happens.
     
    Last edited: Aug 2, 2020
  6. Biodegraded

    Biodegraded Friend

    Pyrate Contributor
    Joined:
    May 28, 2017
    Likes Received:
    7,982
    Trophy Points:
    113
    Location:
    Vancouver BC
    If tab delimited would result in two separate columns on each file's github page, that'd make it easier when downloading individual measurements. Tab is just as good for REW import.

    Additional data request: impedance vs frequency profiles for any that are 'interesting' (say 3x or more difference between trough & crest), for calculating and displaying the effect of amp output impedance. If you want to put this calculator or a link to it on the page too, that's fine with me - you might want to check it first though!

    https://www.dropbox.com/s/cxjar2y10...plifier output impedance calculator.xlsx?dl=0
     
  7. Bill-P

    Bill-P Level 42 Mad Wizard

    Pyrate
    Joined:
    Sep 25, 2015
    Likes Received:
    4,801
    Trophy Points:
    113
    Location:
    San Jose, CA
    Oh, that's cool!

    Then I guess it's time I reveal something I worked on just as a proof of concept:
    http://billp.site/projects/chart-tool/

    That's a headphone frequency response comparison tool I built for the web a while back. If you add one of the files in the Github onto that tool, you can add about +90dB offset to visualize it with the default parameters. You can then hover your cursor to see more detailed data. Please try it.

    P.S.: feedbacks are welcome! I haven't touched that for over a year by now... but if you guys need more features or want the UI to be better, just let me know.

    P.S. 2: I'm guessing someone will suggest that the tool take in something like a "compensation curve" and apply it. Well, will think about that.

    Edit: quick usage guide:

    1. Navigate to the URL
    2. Click on "Add Data", and choose a data file on your computer
    3. You'll see a new row added with the filename, click on the box that says "0.0", type in "90.0" and click "+Offset". Enjoy!

    I provided my old measurement data of LCD-2 and HD600 as a point of comparison.
     
  8. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,777
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    Very cool. Is REW format space or tab delimited? I can always do an FRD format, but without the phase info. The FRD format may be a good idea.
     
  9. Bill-P

    Bill-P Level 42 Mad Wizard

    Pyrate
    Joined:
    Sep 25, 2015
    Likes Received:
    4,801
    Trophy Points:
    113
    Location:
    San Jose, CA
    I think by default, it's space delimited. I can tweak the tool to auto-detect for more delimiter formats if that'll make life easier.
     
  10. Biodegraded

    Biodegraded Friend

    Pyrate Contributor
    Joined:
    May 28, 2017
    Likes Received:
    7,982
    Trophy Points:
    113
    Location:
    Vancouver BC
    REW can be set to space, comma, tab, semicolon for export, and for importing will recognize any of those.
     
  11. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,777
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    I'm going to set to tab character and re-up everything.
     
  12. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,777
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    @Bill-P: Your chart program likes space delimiter. Maybe I should revert to space.

    Anyway, it does work right now, at least with the space delimiter:
    upload_2020-8-2_17-32-12.png

    Let me reconvert everything back from ascii 9 to ascii 32 and re-upload. :D
     
  13. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,777
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    Works beautifully now.
    upload_2020-8-2_17-42-41.png
     
  14. zerodeefex

    zerodeefex SBAF's Imelda Marcos

    Staff Member Pyrate BWC
    Joined:
    Sep 25, 2015
    Likes Received:
    14,051
    Trophy Points:
    113
    @Bill-P should we add your tool to the repo? The intent is to blow open the doors on making all of this as accessible as possible.
     
  15. Bill-P

    Bill-P Level 42 Mad Wizard

    Pyrate
    Joined:
    Sep 25, 2015
    Likes Received:
    4,801
    Trophy Points:
    113
    Location:
    San Jose, CA
    Oh, please feel free to. But if you need the URL to be more elegant or something like that, please let me know. I've been meaning on expanding on it for a year now... but work and other stuffs have prevented me from working more on it.
     
  16. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,777
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    He meant put the code in the Git repository so others can branch or fork your code with improvements, functionality, etc. I would need to set you up. You can be set up as the owner so you can then "pull" in desirable branches. Open source community development.
     
  17. Bill-P

    Bill-P Level 42 Mad Wizard

    Pyrate
    Joined:
    Sep 25, 2015
    Likes Received:
    4,801
    Trophy Points:
    113
    Location:
    San Jose, CA
    Oh, yeah, sure! That'd be awesome.

    That'd also make it easier for me if people can contribute to the code with features that they think would be useful.
     
  18. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,777
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    I re-upped some files (used wrong side calibration for them by accident) and uploaded the remaining EARS3.2 (no screws on the coupler) measurements. There isn't that many of them. Next step will be to up the EARS3.0 measurements (the ones taken with the screws) and then the flat plate coupler.

    I may write some simple Windows based visualization tools and put the source and binaries up in a separate code / tools repository. A ton of stuff to do. @Bill-P's web doodle is actually quite nice. I've been using it myself to spot check the data. Thanks Bill!
     
  19. Hands

    Hands Overzealous Auto Flusher - Measurbator

    Staff Member Pyrate MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    12,285
    Trophy Points:
    113
    Location:
    Colorado
    Home Page:
    It would be killer to leverage some of the tools you've built, if nothing else for consistency. And because I end up doing all visualization in Excel, which sucks from a labor perspective.
     
  20. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,777
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    Uploaded over a dozen measurements cEAR30 (the MiniDSP EARS with the screws still on them). These may be a little less consistent with the seal because of the screws. Actually I went back and did not upload the ones that looked funny. The compensations on these were tweaked so they would align with the earlier cEAR32 (the EARS without the screws) for the sake of consistency. It won't be perfect though because of seal and placement inconsistencies, human error, alignment of the stars.

    The next phase will be some analysis of the flate plate coupler data and refinement of the compensation so that it aligns with the EARS. This could be more difficult since the the FPC does not have a pinna. However, where datasets for both exist, it maybe interesting to see which headphones have more interaction with the pinna and which do not.

    However, there is a ton of FPC data.

    http://hpdb.io
     

Share This Page