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

Dirac for Apple Earbuds

Discussion in 'Sound Science' started by bigshot, Nov 13, 2012.
  1. bigshot
    Has anyone tried this yet?

    Dirac HD Player Lite by Dirac

    In theory, this could work. I'd be interested in hearing from people who try it. They could create DSP correction signatures for specific brands and models of headphones and save us all a lot of money.
  2. chewy4
    Eh, I can't see them expanding much beyond popular headphones like earbuds and Beats. It would be interesting to try if they did though.
  3. Wembely
    What you are looking for already exists and its great.  its the goldenears.net app called Accudio.  it's the best $4.99 I ever spent on music quality and  making IEM and headphones sound amazing.  But beware it can make some cheap IEM sound better than multi-hundred dollar flagships - so may induce some buyers regret once you see what 5 dollars worth of software can do to your transducer hardware.  no kidding!
    Find it here: https://itunes.apple.com/us/app/accudio-pro/id553759905?mt=8  don't worry about the reviews - most are complaining about it not having features it never claimed to have.  as far as a DSP - it is brilliant and hugely effective
    I have no affiliation with these guys other than being a satisfied customer.
  4. tumburu
    I tried the Accudio with my k3003 and the sound is not realistic at all, it basically transforms it into a mid centric phone. Sure, you can tweak a bit the sound afterwards, but it's far from what I'd like to hear, plus it adds big gaps between the tracks. I believe the "curve" goldenears.net proposes represents their view on "neutrality", but as an audio engineer, who puts neutrality above other qualities in his listening chain, I can tell you, the result is not what I expect.
    The Dirac player OTOH is pure fun. I think everybody that has a pair of apple buds should have it, can't believe it's free.
  5. xnor
    K3003 has a very v-shaped frequency response. I have not seen the correction curve for this earphone, but if it's anything like the others the result should be fairly flat.
  6. tumburu
    We certainly hear things different. To me the k3003 have a bright response (most energy in the treble area of the spectrum).
  7. ultrabike
    That qualifies as a v-shaped frequency response if the bass is also emphasized. The mid-range is what gets attenuated relative to the whole spectrum in a v-shaped FR headphone. Consider the DT-990 which has quite a bit of bass, and even more tremble at the expense of the mid-range.
    That said, like the OP, I'm more interested in impressions related to the performance of the Dirac app. Have you tried it?
  8. tumburu
    IME the bass on the akgs is rather neutral, even the goldenears.net "curve" shows an elevated bass response is needed for a headphone to be perceived as having enough bass. Sometimes I'd say k3003 have a decent bass, definitely not emphasized, compared to the really neutral monitors (Genelec, Klein Hummel and Barefoot) that I use in the studio. Sony EX1000 have a more speaker like bass to me (slightly more bass quantity).
    Back to the Dirac app, I'm really enjoying it. It made me use again the white earbuds, which is quite something, considering I never used them for anything than phone calls in the last year. I'd like to see more popular (non Apple) models added (like ie8/80 or gr07 etc), some general tone controls (it can get too bright sometimes), gapless playback and flac support :D
    One really basic function that it missing is the inability to fast forward/rewind through a track from the remote.
  9. ultrabike
    Thanks tumburu. It seems there still is some room for improvement (format support and basic controls), but brightness issues aside, did you experience any clipping or flagrant distortion issues?
  10. tumburu
    No, these guys are clearly pros. The headroom is created by lowering the level of the signal (much like the Accudio app), so you'll need to crank the level up more than before.
  11. applehead
    I'm listening to the Dirac app with earpods now, it sounds pretty good. Compared to the stock player, the mids are boosted quite a lot, with the highs having a slight boost and the bass remaining the same. Subjectively, the vocals on most songs just seems to be a lot clearer. In stock form the earpods just felt way too bassy for me, with the bass leaking into the midrange. While the Dirac app fixes that somewhat, I stll feel the earpods are too bassy but the app prevents the bass from being too overbearing.

    If I were to guess about the exact frquency ranges affected, I think there's a hump at around 500-6khz,with the peak being at 1khz leading to a gradual decline.
  12. yuriv
    I'm sure someone would have posted something like this eventually. Even the much-maligned RMAA can tell us a story:
    As far as I can tell, Dirac isn't doing things like crossfeed or adding delays or reverb. Crossfeed would have been nice, but IMO it's better not to mangle the signal like Aphex and BBE do.
    The correction filters for the Earpods seem to be similar. Accudio is more conservative and it uses fewer bands of parametric EQ. They say that they try not to boost nulls caused by comb filtering and and that they don't try to cancel the sharp peaks caused by resonance. Those half-wavelength resonances that come with IEMs are usually dependent on things like tip selection and insertion depth, so the correction filters for those things are guesses, at best.
    But when I played with the Earpods and a tone generator, I found the peak at 6 kHz to be largely unaffected by the looseness or tightness of the fit. The data on goldenears.net seems to support this. So the decision by Dirac to null the response at 6k works for me. But I'm only one data point. YMMV. The frequency response charts on goldenears.net use 1/3-octave averaging, so we see shorter and wider peaks than if they had reported the raw response. Accudio's filters seem to be correcting the averaged measurement.
    The Earpods' frequency response at the low end really depends on how and where you have them positioned, and both Accudio and Dirac are making guesses on the kind of fit that you have. So the settings on both programs won't work for everyone.
    The big difference I see on both graphs is the response at 3-4 kHz. Accudio wants it to be around 4 dB higher than Dirac. From what I can hear with music and tone generators, Accudio is more in agreement with any of several sets of of my speaker rigs, all of which measure flat at the listening position (before applying a so-called X-curve that rolls off the treble gently).
    Here's what Accudio says it's doing (seems honest enough):
    I suppose, if you really wanted to, you could try your best to copy Dirac's frequency response using Equalizer or custom mode in Accudio. The results might sound very, very similar.
  13. stv014
    Does it try to correct the very high bass distortion of the Apple ear buds (that would not be easy, though, and would require knowing the actual output level on the ear buds) ? If not, then it would be easy to copy the impulse response, although it may not be allowed for copyright reasons.
  14. firev1
    Wish purrin can just do a CSD of accudio and Dirac just to see. Really wonder if the Dirac filter is really helping with the time domain as it claims to do. Yuriv, subjectively how do Dirac and Accudio compare? Especially regarding any imaging or time domain related phenomenal?
  15. yuriv
    Quote: by stv014
    I'm pretty sure it doesn't try to correct the distortion on the earbuds. That's not easy, even if you know the output level. If you don't have feedback, the way to reduce the nonlinear distortion in a system that distorts is to feed it a pre-distorted signal. Two wrongs make a right, if you will. Dirac plays back a 60 Hz sine wave cleanly, even at maximum volume; 2nd harmonic is around 90 dB down on an iPhone. Also, Dirac doesn't change the shape of the impulse when you change the volume level, but that's to be expected since the FR didn't change with level either. Some screenshots of impulse response measurements follow.
    Maybe Dirac really is doing something besides being a straight EQ. Maybe not. Here's what I get out of an iPhone 4S when I play back an impulse. It's harder to tell what's going on because of the iPhone's minimum phase filtering. Straight from a digital output would be better.
    Accudio - Earpods
    Accudio - Flat -7dB
    Dirac - Earpods
    Dirac - Effect off

    Quote: by firev1
    You can derive a CSD from the impulse response. I haven't played with the data yet, though. The CSD that you really want to see is the one that comes out of the acoustic measurements of the Earpods, because those are the ones that ultimately count. But maybe it's time to fire up more software so we can figure out what it's really doing. Subjectively, I think Dirac's decision to null the response at 6kHz works. Excessive energy in the presence region can sound harsh sometimes. But that might not work for everyone. You can get Equalizer or Accudio to do that when on manual.
    As far as imaging goes, I'm not the right person to ask. Headphone imaging to me, just sounds wrong. The less attention a headphone calls to its imaging, the better--unless it's playing back a binaural recording. Sometimes crossfeed can help reduce that annoyance. So I'm not terribly impressed by the so-called "soundstage" of things like an AD700 or a K701. I'd love to hear the Smyth Realiser one day. As far as I can tell, Dirac doesn't add a delayed, frequency-dependent-attenuated copy of the signal from one channel to the other.
    Maybe someone could copy Dirac's frequency response using another EQ, then compare the sound, and maybe look at the resulting impulse response also.
    tktran, ultrabike and MarcadoStalker7 like this.

Share This Page