Vienna Symphonic Library Forum
Forum Statistics

182,112 users have contributed to 42,205 threads and 254,666 posts.

In the past 24 hours, we have 4 new thread(s), 10 new post(s) and 68 new user(s).

  • Higher sample rates

    Recording guitar in my studio sounds a lot more detailed at 88.2k than 44.1k, so that is my preferred sample rate. I then downsample to 44.1k. I'm getting a lot of clicks with not that many instances of VIs running. When I have say about 5 instances, the G5 (2.5GHz dual processor 4 Gb RAM) gets really slow and lags behind my key commands slightly. Very annoying. Is this because my sample rate is set higher than the 44.1k VSL samples are at? I would hate to be limited to record guitar at 44.1k.

  • this is kind of *foreign territory* for me here ;-) however you didn't mention bit depth ... of course when using another sample rate some conversion takes place *on the fly* which needs CPU - how much depends on the used application.

    out of interest: isn't the 88.2 somehow a dead end meanwhile?

    christian


    and remember: only a CRAY can run an endless loop in just three seconds.
  • Hi Christian, I believe the reason to sample rate at 88.2 is that the conversion to 44.1 is a simpler and possibly more transparent process due to the maths. In the same way if your going to ultimately output 48k (video soundtracks) then 96k would be a better way to go. i seem to remember that the VSL samples were recorded in parallel sample rates so no conversion happened? Another point is that the quality of the convertors will usually transcend the sample rate used - top range pro convertors recording or playing back 44.1k sample rates will sound better than budget soundcards running at 88.2 or 96 or even 192. Julian .....sent from my safari (browser!)

  • All our samples are recorded in 96 kHz / 24 bit, stored in 32 bit Floating Point, and are converted to the release format (today this is 44.1 kHz, 24 bit) during the last step in the production cycle, prior to the mapping for the Vienna Instrument. The conversion is done on behalf of a proprietary sampling rate converter which was developed especially for our needs and quality standards.

    BTW - in the case of _good_ SRC it doesn't matter if you use simple multiples (44.1 / 88.2) or not (44.1 / 96) - the benefit resulting from simple ratios is a typical Urban Legend. The main issue for our ears is the anti aliasing filter.

    HTH,

    /Dietz - Vienna Symphonic Library

    /Dietz - Vienna Symphonic Library
  • Hi jrm1,

    I kind of read your question from another perspective. 

    Those clicks; I believe this has to do with your buffer size.

    You should increase the buffer size in your DAW or software.

    What soft are you using?


  • Reading the posts it looks like the most of us is working for video.

    Knowing that 44.1 is bit too tight for the Nyquist-Shannon teorema, and it was choosen because of the early CD tecnology,

    Why don't you use the 48K sampling rate as the release format. That's the same as DAT and other more professional music products

    Sergino


  • In principle you're right, Sergino. But many of us work for pure musical CD-releases too, so 44.1 is a valid format nontheless.

    The reason we decided for 44.1 is the simple fact that you don't gain a lot of quality when keeping things at 48 kHz, but you lose 10% of performance. ;-)

    Kind regards,

    /Dietz

    /Dietz - Vienna Symphonic Library
  • last edited
    last edited

    @Dietz said:

    In principle you're right, Sergino. But many of us work for pure musical CD-releases too, so 44.1 is a valid format nontheless.

    The reason we decided for 44.1 is the simple fact that you don't gain a lot of quality when keeping things at 48 kHz, but you lose 10% of performance. 😉

    Kind regards,

    /Dietz
    That's not what my dog thinks!! Julian

  • ... hmmmm .... you are right, how could we forget about this very important target-group ... tstststs .... 8-)

    **********

    Happy New Year!

    **********

    /Dietz - Vienna Symphonic Library
  • Not sure why you think of 88.2 as a dead end. As I said, the resolution in a recording of a hardware synth or guitar is better than at a lower rate. Not a dead end at all. I then convert to 44.1 or 48 depending on your delivery needs. I'd like to confine the debate to my original question, which concerned the combination of VSL with guitar/synths etc. What sample rate is best for running VSL? 44, 48, 88.2? The G5 is, I understand, less stressed by running a project at 44.1, but by how much? I've tried different buffer sizes but that hasn't made much appreciable difference.

  • In short: 44.1 makes most sense.

    HTH, /Dietz

    /Dietz - Vienna Symphonic Library
  • I'm sorry to hijack the tread a bit here, but I was wondering Dietz, if you could answer this as I am confused about this thread.

    I understand how this can be true: "Something recorded at 88.2khz played back at 88.2khz will sound better than something recorded at 44.1khz played back at 44.1khz."

    However, I DO NOT understand how THIS can be true: "Something recorded at 88.2khz played back at 44.1khz will sound better than something recorded at 44.1khz played back at 44.1khz."

    If jrm1 is recording his guitar at 88.2khz and ultimately samples down to 44.1khz for the finale mix I don't see how recording this way would make it sound better in the 44.1khz mix. I would think the only benefit to recording at 88.2khz is the freedom to go back and sample down at various rates, or if you have a playback device that outputs at 88.2khz.

    Thanks, Brian 


  • I meant to say that the VIs _themselves _won't sound better at any other SR than 44.1 . I was not referring to jrm1's guitar recording.

    It is quite possible that processing applied on the signal coming from a VI will sound (a bit) better due to the higher resolution of EQs and dynamic processors.

    HTH,

    /Dietz

    /Dietz - Vienna Symphonic Library
  • "how could we forget about this very important target-group" hahahaha

  • last edited
    last edited

    @jrm1 said:

    I'm getting a lot of clicks with not that many instances of VIs running. When I have say about 5 instances, the G5 (2.5GHz dual processor 4 Gb RAM) gets really slow and lags behind my key commands slightly. Very annoying.

    In addition to the pros and cons of 88.2 vs 44.1 already mentioned, putting in another 2GB of RAM will help in that computer. For running intensive VIs these days, 4GB is the new minimum, imho. Now that VSL can access its own RAM outside of the DAW, there's almost no reason not to give your computer the "real estate" it needs. G5 RAM is a lot cheaper than it used to be.

    Running the VIs from a different drive than your system drive or your audio/project drive also helps when drive busses are not shared. Firewire remains a problem on PPC as it just doesn't perform as well as SATA or eSATA. Checking settings in your DAW that govern how CPU cycles are allocated may require some fine tuning. This includes buffer sizes, any work priorities, pre-rendering, and disk usage settings. Each DAW is a little different, but the concept is the same.