• 3 Posts
  • 62 Comments
Joined 1 year ago
cake
Cake day: July 16th, 2023

help-circle






  • Are these families somehow more meritorious than the rest of the population?

    lacking multi-generational connections is still a pretty rosy picture of disadvantage. Statistically “unmeritorious” parents are far more likely to have their child suffer from malnutrition due to lack of money and neglect due to the parents working 2 jobs or having substance abuse issues. If the country has private schools, they won’t have access to them and due to living in a low-wealth area their public schools will have a disproportionately high amount of other neglected and abused kids which makes everything harder.








  • the direct chain I can see is

    “can you string words to form a valid RSA key”

    “I would hope so, [xkcd about password strength]”

    “words are the least secure way to generate random bytes”

    “Good luck remembering random bytes. That infographic is about memorable passwords.”

    “You memorize your RSA keys?”

    so between comments 2 and 3 and 4 I’d say it soundly went past the handcrafted RSA key stuff.



  • sus@programming.devtolinuxmemes@lemmy.world-----BEGIN PRIVATE KEY-----
    link
    fedilink
    arrow-up
    2
    arrow-down
    1
    ·
    edit-2
    4 months ago

    if you know there are exactly two additional characters

    this is pretty much irrelevant, as the amount of passwords with n+1 random characters is going to be exponentially higher than ones with n random characters. Any decent password cracker is going to try the 30x smaller set before doing the bigger set

    and you know they are at the end of the string

    that knowledge is worth like 2 bits at most, unless the characters are in the middle of a word which is probably even harder to remember

    if you know there are exactly two additional characters and you know they are at the end of the string, the first number is really slightly bigger (like 11 times)

    even if you assume the random characters are chosen from a large set, say 256 characters, you’d still get the 4-word one as over 50 times more. Far more likely is that it’s a regular human following one of those “you must have x numbers and y special characters” rules which would reduce it to something like 1234567890!?<^>@$%&±() which is going to be less than 30 characters

    and even if they end up roughly equal in quessing difficulty, it is still far easier to remember the 4 random words





  • sus@programming.devtomemes@lemmy.worldweird looking gear
    link
    fedilink
    arrow-up
    1
    ·
    edit-2
    4 months ago

    step 1. Try presets that have already been calibrated to some target for those specific headphones. There are hundreds to thousands of headphones included in the bigger preset collections.

    step 2. tweak the EQ values by yourself by ear if you want to. There is no objectively best sound, so it comes down to your personal preference anyways, and you can’t measure that in any practical way (and I’d say neither can the companies making expensive headphones, which is why there are hundreds of different headphones both cheap and expensive with different frequency responses and more getting made all the time)


  • sus@programming.devtomemes@lemmy.worldweird looking gear
    link
    fedilink
    arrow-up
    1
    ·
    edit-2
    4 months ago

    “tonality characteristics” and “soundstage” are subjective words that have no concrete definition. Other similar words are “grain”, “speed”, “separation”, “resolution”. They can’t be objectively measured, and are most likely just another function of frequency response.

    The differences between headphones are most likely your ear having a different shape from the reference ear used to make the eq targets, leading to a different final perceived frequency response. (or limitations in the accuracy of the measurements, most targets I believe are “smoothed”)

    I’m going to trust the (claimed, who knows, maybe oratory1990 is a liar) consensus of audio engineers over your anecdotes. As I said there are plenty of audiophiles whose “lived experience” is that $2000 golden cables are necessary and that they can tell the difference between any $200 and $1000 DAC (even though a decent DAC in that price range already has a dynamic range and signal-to-noise ratio of 100-120dB which should be totally indistinguishable from perfectly clear audio for all humans

    personally the only decent-ish headphones I have are DT 880 600 ohm and a JBL 760NC. The latter kind of fills all the boxes of being a wireless headphone and has poor reviews and a poor default sound profile. But after EQing both, I can’t really notice any difference except when very carefully doing side-to-side comparisons (besides the much better comfort of around-ears vs over-ear).

    In contrast I believe I can tell, with some songs, the difference between 320kbps mp3 and flac (just 44.1khz), but even there I’m not sure it’s not just placebo

    Usability is kind of secondary, android should have jamesDSP and the venn diagram of people that know the best headphones to buy (instead of beats by dre) and who can setup an EQ (install an app and follow written instructions) should have a lot of overlap

    I will say though that more expensive headphones are probably going to last longer and are probably much more comfortable


  • sus@programming.devtomemes@lemmy.worldweird looking gear
    link
    fedilink
    arrow-up
    2
    arrow-down
    1
    ·
    edit-2
    4 months ago

    The thing is, distortion (maybe more accurately called nonlinearity) is the only known objective way to measure the difference in sound quality between two headphones EQ’d to the same target. (there are some other measures like signal-to-noise ratio but they are even more useless) And the difference in the value becomes very small for a technically good $50 headphone and the best headphone ever made. (technically good eg. the natural frequency response isn’t crazy far from your target and the nonlinearities are competitively low)

    Now, two headphones EQ’d to the same target, even if both are measured to result in the exact same sound, won’t actually sound the same to your ears because the “head dummy” used for the test doesn’t have the same ear shape and characteristics as you do. But unless there is some strong evidence that the headphone manufacturer has a better methodology than what is publicly available, then there’s no reason to think they are somehow able to account for your specific ear’s needs without custom designing the product just for you. - You’re left with having to either EQ yourself, or using dozens of headphones and testing which you like the most. And the EQ route is going to be much faster and cheaper

    for sources, these discussion seem the most useful

    https://www.reddit.com/r/headphones/comments/144yaiq/why_dont_we_measure_headphone_resolution/jni4z70/?context=5 (whole thread is useful)

    https://www.audiosciencereview.com/forum/index.php?threads/is-there-any-way-to-objectively-measure-headphone-resolution.17684/

    you can say that most people who spend a lot of time and money trying to achieve “perfect audio” seem to think that EQ is only a supplement to already good headphones, but given that there has been no success at objective measurements of quality and that many people swear the thousands they spent on insulated golden cables improve their audio quality, I err on the side of saving my money.