Announcement

Collapse
No announcement yet.

Sample Rate vs Filters , Part Two

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Sample Rate vs Filters , Part Two

    Group,

    We had a long discussion regarding the optimum sample rate and settings for LP records. Many subjects were covered and I had a number of comments/observations about my experience of 96 kHz vs 44.1 kHz , and other things.

    With my recent upgrade to a "decent" computer system, I'm re-exampling my earlier work. One reason is that the computer that I used in the past did NOT sample correctly and gave me false info. So.....

    I'm trying flat vs RIAA preamps and 96 kHz sampling vs 44.1 kHz sampling last night on a 78 record "House of Blue Lights" by Freddie Slack and Ella Morse (great stuff!!).

    My multifilter for 78 records was orginally built up using my old system with 44.1 kHz sampling.

    What I found was that the best noise reduction (clicks and hiss) was obtained by recording at 96 kHz, then converting to 44.1 kHz, and then running the multifilter. I used the flat transfer preamp.

    I tried to optimize the multifilter chain for the 96 kHz rate, but I always found that the BEST results were to first, record at 96 kHz, then convert to 44.1 kHz ?

    I tried recording at 44.1 kHz and then running the filters, but in each case, the results were always better if I FIRST started at 96 kHz and then converted down to 44.1 kHz. for the filters.

    So...my "gut" says that more information to the various filters in my multifilter ( Ez Impulse/Median/CNF) should sound better, yet if I use the lower rate (44.1 vs 96), it's better ?

    Is the answer that I just have to optimize all the filters for 96 vs 44.1 or, do the various software routines have an optimum sample rate ?

    Maybe the conversion from 96 to 44.1 by the DC software gives the filters a little more ultrasonic info than a straight recording at 44.1 kHz ?

    Marc

  • #2
    Hi Marc,

    I am a little unclear about what you are saying, but I think that the optimum settings for the filters (multifilter, impulse filter(s) or continuous noise filters) are sample rate dependent. What I gleen from previous discussions on this subject here on the forum is that better results often can be had using 96 kHz compared to 44.1 kHz. But, the caveat would probably be that the best results with either sampling rate are obtained via providing the best settings for the filters at each sample rate. So, presets designed and optimized for files at 44.1 kHz will not be optimal at 96 kHz. It will likely take some re-adjustments of the various parameters.

    Generally, what I think that folks have done is this sequence:

    1. Transfer recording flat at 96 kHz

    2. Use the Virtual Phono Preamplifier in Diamond Cut to provide the proper EQ Curve to the file

    3. Use the Impulse Filter(s) to remove clicks, pops, ticks, etc.

    4. Convert file to 44.1 kHz

    5. Apply the CNF to reduce hiss

    Etc.


    Craig

    ps - I do not recall if folks have claimed superior noise reduction of the CNF at higher sample rates and that is why I placed the file sample rate conversion before the CNF process. Maybe someone who has had experience with that can comment. I can say that the bin widths for a given fft size will roughly double when going from 44.1 kHz to 96 kHz - - - so the selectivity of the bands will be affected for a given set of CNF parameters.
    Last edited by Craig Maier; 05-15-2012, 01:06 PM.
    "Who put orange juice in my orange juice?" - - - William Claude Dukenfield

    Comment


    • #3
      Craig,

      Thanks for the reply. While my posts are clear to me, others can find them "cloudy" !

      Your comment about the CNF at 44.1 kHz explains why my multifilter works better with the approach I stated above. I'll work on updating my Impulse filter settings for 96 kHz.

      Marc

      Comment


      • #4
        Hi Marc,

        While it it true what I said about the selectivity of the CNF bin decreasing with constant fft size but higher sample rates, that does not necessarily imply that the performance will be poorer. When the selectivity is decreased (bin frequency width roughly doubles in the case in point) the transient response of the system improves. So, that is the balancing act with the CNF - selectivity on one side of the scale and transient response on the other.

        Craig
        "Who put orange juice in my orange juice?" - - - William Claude Dukenfield

        Comment

        Working...
        X