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

Upsampling 16/44hz FLAC to .wav 24/96: benefits????

Discussion in 'Sound Science' started by rahzim, Jun 29, 2012.
  1. rahzim
    Hello all,
              I was reading a review on dbpoweramp and noticed some peculiar claims about sample rates.  The author was touting a feature of dbpoweramp that allowed users to convert compressed flac to uncompressed, as if doing his had any impact on audio quality.  Even more questionable than this was some users claims that conversion from 16.1/44hz compressed flac to 24/96hz uncompressed wav files resulted in better quality playback on SOME systems.  The user claimed this was due to some DACs reconstruction filters being inadequate when converting compressed files.  The logic followed that without having to reconstruct uncompressed wav files, the resulting audio quality would benefit.
    Is there any truth to these claims at all? I was hoping some of the more knowledgable users here could provide their insight on this.  The link for this website and the user comments I'm talking about is below. Thanks!
    - Tim
  2. ChipnDalebowl
    No difference...you cannot make 16-bit 44khz audio sound better than it is. Even a conversion to 24-bit 96khz audio still only gives you the original resolution...it can't put in what isn't there.
  3. rahzim
    Right, but the argument wasn't about adding resolution that isn't there.  They were arguing that since some DAC's reconstruction filter isn't adequate, the resulting conversion from a compressed file would be of lesser fidelity than reconstructing an uncompressed .wav file.
  4. grokit
    24/96 will add digital headroom, which helps to prevent distortion when using DSP.
  5. xnor
    24 bits will add headroom, but this can be done during playback. All you need is to check the playback devices' options.
    Resampling to 96 kHz imho doesn't make much sense since DSP plugins that need it will automatically oversample. Some DACs however could benefit from resampling, e.g. those with massive treble roll-off, NOS ... but I'd just get a properly implemented DAC instead.
  6. grokit
    Agreed, but for myself upsampling is always done during playback, the choice is whether to do it with hardware or with software.
    I think that what you are discussing is upconversion.
    Just semantics though, we know what we're talking about!
  7. mikeaj
    nevermind the thread moved too fast while I was away...
    You don't need the file to have 24-bit data to run the DAC in 24-bit mode, so even if you need headroom for digital volume controls or other processing, you don't need the actual files to have 24-bit resolution.
    By the time the data gets sent to the DAC, it's already been uncompressed, so whether or not the data is stored in a (lossless) compressed format or not, it doesn't matter.  You literally are sending the exact same data to the DAC either way, so how is the sound going to be any different?  Maybe the DAC takes a peek inside your hard drive and sets itself to low-performance mode if you are using any data compression.  Maybe DACs don't like zip files or the Internet either.  CPU usage for realtime decoding of the audio data should be very low and not matter, though sure, I guess theoretically in some freak scenario you could have the higher CPU usage combined with some other programs running, keep the system busy enough that maybe some audio data doesn't get handled in time (so a brief click / dropout) or some other kind of extremely fringe scenario.
    However, there is maybe one valid point.  Some DACs are really poorly designed / improperly configured (some of the "audiophile" offerings, some of the eBay specials, and so on), and may legitimately perform worse at 44.1 kHz sampling rate than at something else.  In that scenario, if your equipment blows, then resampling the audio to a sample rate it can handle better, may improve performance.  In general you want to avoid resampling and all the minor issues associated with that, but it could be the lesser of two evils sometimes.
  8. xnor
    Here's some definitions:
    Increasing sample rate:
    Upsampling = inserting zero-valued samples between original samples
    Interpolation = upsampling followed by filtering to remove the undesired spectral images
    Decreasing sample rate:
    Downsampling = throwing away samples
    Decimation = filtering followed by downsampling
    Resampling = combining interpolation and decimation, for example: 44100 Hz * 320 / 147 = 96000 Hz (note the fractional resampling factor).
    Sigma-delta DACs use an interpolation filter in front of the modulator.
  9. ultrabike
    I read the articles (including the linked http://www.audiostream.com/content/cut-flac) and they had my brain going. The author, seems to allude to the possibility that the losslessly compressed audio stream might suffer (vs the uncompressed stream) since the decoding of the lossless stream might tax the system (maybe heat the components or create some internal interference) ... and possibly generate HW noise? I don't know. If noise is the problem, probably best to upgrade your audio card or computer. I can see an advantage though: it may take less time to do a wav to uncompressed flac transfer (large library), but you obviously will pay the price in storage.
    Also, performing upsampling will necessitate anti-aliasing filters (interpolation) as discussed here. These filters are not lossless operators. Different filters have different trade-offs. The results will also be different depending on the DAC, Amp, Headphone, etc. all of which are also not lossless operations either. I see a valid argument in saying that going to 96kHz alleviates the DAC and reconstruction filter requirements. However, I'm having a hard time reconciling the notion that lossless FLAC compression is an issue, but upsampling is non-issue and much better. I feel FLAC decoding in most cases would be a non-issue, but upsampling will have a significant impact (positive or negative) depending on your HW.
    EDIT: After re-reading the articles, the author does not offer an explanation for his FLAC vs WAV perceived performance difference, the readers in the comment sections do. I'm not sure what to make of those articles at this point.

Share This Page