Pre and Post Effin Ringing and shit like that

Discussion in 'Blind Testing and Psychoacoustics' started by ultrabike, Aug 17, 2016.

  1. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,960
    Trophy Points:
    113
    Location:
    Irvine CA
    You are thinking of all this completely and thoroughly wrong.

    If I have a recording with nothing, and I do mean nothing, but an impulse, then (in a nutshell) implied is ringing between samples.

    When the system reconstruct the signal in the analog domain, the analog reconstruction filters interpolate between samples. "Ringing" will happen between the samples due to the bandlimited analog interpolation.

    For a signal to be a Dirac impulse, it has to have infinite power and infinite bandwidth.

    SUCH THING. DOES NOT EXIST.
     
    Last edited: Jan 21, 2019
  2. MF_Kitten

    MF_Kitten Banned per own request

    Banned
    Joined:
    Jan 1, 2016
    Likes Received:
    181
    Trophy Points:
    43
    See, that tells me A LOT right there! Thanks!
     
  3. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,960
    Trophy Points:
    113
    Location:
    Irvine CA
    Think about it this way: Nothing is infinite power and bandwidth, which are requirements to truly attain delta Dirac outputs.

    Even the conversion, if we were to remove the analog filters is very band-limited and power limited. Many implementations output a staircase approximation (S/H), as opposed to a unicorn impulse train. A stair case approximation implies a sinc function low-pass frequency response. Some implementations use pre-emphasis to compensate for the sinc function roll-off and then apply digital and analog filters to remove the high frequency lobes in the frequency response in order to achieve a flat response in the passband and high attenuation in the stopband.

    If we were to focus completely on the time domain ringing, one could keep a flat passband and control the time domain ringing to some extent by oversampling and using some less step and controlled roll-off on the filter (similar to windowing in the other domain). I don't think that necessarily buys you much in the passband though. In such a case, one maybe trading transition band for less "ringing", and hopefully that transition band is above 20 kHz.
     
    Last edited: Jan 22, 2019

Share This Page