Vienna Symphonic Library Forum
Forum Statistics

194,336 users have contributed to 42,916 threads and 257,955 posts.

In the past 24 hours, we have 2 new thread(s), 12 new post(s) and 82 new user(s).

  1. Home
  2. Software
  3. Vienna MIR & Vienna Suite
  4. Does Mir Pro reverb sound better at higher sampling rates?

  • Does Mir Pro reverb sound better at higher sampling rates?

    A
    A
    Over the years I've heard many people say reverb tails sound more natural/deeper when they work at higher sample rates, i.e. 96khz instead of 44khz. Would this only be true of algorithmic reverbs like Lexicon? Or would it also be true of convolution reverbs like Mir Pro? Are Mir Pro's source impulse responses only 44khz, and then upsampled when working at 96khz? If so, then there would not be any difference in quality working at 96khz, right? Just curious :)
  • DietzD
    VSL Team
    DietzD

    The short answer would be: No, not really. :-)

    The long answer:

    MIR's impulse responses are recorded in 96kHz, 24bit. They are stored and processed in 96kHz, 32bit floating point. In a final step they are converted to 44.1kHz before mapping and delivery, by means of VSL's proprietary sample rate converter tool (which is also used for all sample libraries).

    During operation, the IRs are converted by the MIR engine to any chosen sample rate. So, as far as MIR Pro is concerned, there will be no difference in sound quality (... which doesn't mean that other components of your system _couldn't_ benefit from higher SRC - or OTOH behave badly due to the higher workload ... your mileage may vary).

    The reasons why we record in a different format than we deliver are manifold:

    - Our measurements have shown that the reverb tails of real rooms have little to no spectral content above 20 kHz (except some very special cases). Many orchestral venues show almost no reflections above 12 or 15 kHz, especially from a listener's perspective. Sampling rates above 44.1are meaningless, in that case (although they may be helpful during pre-processing).

    - Higher sampling rates are mostly useful to avoid the artifacts (ringing) of hard anti-aliasing filtering during recording; higher sampling rates allow for smoother filters. The off-line re-sampling tool mentioned above assures perfect quality in that respect.

    - Within MIR, the dry signal components themselves won't be affected by the IR's sample rate anyway, so that's another point which can be neglected with quiet conscience. :-)

    - The higher the sampling rate, the higher disk storage space, loading times, positional pre-rendering times, RAM demands and CPU consumption will be. Especially the latter would be a big problem in case of a massive convolution engine like MIR.

    ... I hope that answers your questions. :-)

    /Dietz - Vienna Symphonic Library
  • A
    A
    Great thorough answer Dietz! Thanks for confirming all that. Such great customer service from VSL.
  • DietzD
    VSL Team
    DietzD

    A personal sidenote (just coming back from the infamous GearSlutz forum, reading a few of those wonderful meaningless threads on the importance of sampling rates higher than 44.1kHz ):


    Being a sound engineer by profession since about 25 years now, I started working with digital audio recorders around 1990. During all those years I didn't use sampling rates higher than 44.1 or 48kHz (when mixing for video) in more than maybe one percent of the productions I was involved with. This didn't hinder me from finishing all of them. I can't remember a single case of a client complaining "Oh! That's a 44.1kHz mix, isn't it? I can't stand it."


    ;-) 

    /Dietz - Vienna Symphonic Library