Recording using spdif...worth it??

    AFAIK, it's been well-and-long established that plenty of folks can hear the difference between 44.1 and 48kHz, Dean, so you're definitely not alone, mate, as I think you know.


    48kHz vs higher rates is a different story of course, and it seems to me that only those with both excellent ears and high-end monitoring setups claim to notice a difference, and not all of them at that - this could be due to inconsistencies in convertor performance across brands and sample rates (they all have their sweet spots). They talk about imaging and air...


    IMHO, 48kHz is the sweet spot; it's a situation of rapidly-diminishing returns beyond this point.


    All this said, there's nothing wrong with 44.1kHz as long as your convertors are great-quality; it's just that you're likely to eek out that smidgen more of high-end clarity at 48kHz. IMHO, this minuscule performance bump is very-likely to please only us; the consumer will be none-the-wiser.

    There is no theoretical quality improvement by going from 44.1 to 48. But many interfaces are made to work at 48 khz and some can induce problems when doing the rate conversion.

    Quote from vjau75

    There is no theoretical quality improvement by going from 44.1 to 48

    There is. The brick-wall LPF can be set at a higher frequency, thus allowing more-accurate representation of super-high frequencies as well as reducing audible artefacts of the filter.

    Quote from Monkey_Man

    The brick-wall LPF can be set at a higher frequency, thus allowing more-accurate representation of super-high frequencies as well as reducing audible artefacts of the filter.

    What are super-high frequencies ? Frequencies outside of the audible spectrum ? In this case, why should it matter ?

    Here'a a recent article from mixing/mastering engineer Ronan Chris Murphy, who mixed King Crimson, Steve Morse, and many others:


    What Sample Rate do Pros use?


    Bottom line: most pros use either 44.1k (when mixing for CD) or 48k (when mixing for video), higher sample rates are more popular among amateurs, according to his poll.




    Here's an often linked white paper (pdf) by Dan Lavry, who builds high quality A/D converters. It covers sampling theory, and explains in detail why sampling rates higher than 96k are actually detrimental for audio quality.





    And, for good measure, the (also often linked) video from the FLAC inventors at Xiph. It has a very good explanation of sampling basics:


    External Content www.youtube.com
    Content embedded from external sources will not be displayed without your consent.
    Through the activation of external content, you agree that personal data may be transferred to third party platforms. We have provided more information on this in our privacy policy.

    Quote from vjau75

    What are super-high frequencies ? Frequencies outside of the audible spectrum ? In this case, why should it matter ?

    I was thinking mainly of the integrity of HF bell curves' (when EQ'ing) shape being maintained (lower brick-wall filter frequency means the bell shape becomes "chopped off" on one side) on through to a higher frequency.


    I was also thinking of the fact that many peeps I've spoken to say they can definitely hear a difference in their projects and therefore made the switch accordingly.


    Lastly, the artefacts of the steep-but-necessary brick-wall filter inherent in all convertor systems are moved upwards on the frequency spectrum, which essentially reduces their audibility.


    That said, with my substantial HF-hearing loss, it's all theory to me anyway as I cannot verify anything. I've been assured in another thread here just today that I needn't worry about it, so I've received conflicting advice on the matter for some years now.


    The bell-curve-shape thing is real, but I've no hope of hearing any differences.

    Here's a vid from Fab Filter showcasing their Q2 plugin and the "cramping" of bells in typical EQ plugins at lower sample rates, due to the Nyqvist filter cut off, versus the Q2


    (On mobile, fast forward to 4m36s)

    Quote from sambrox


    Here's a vid from Fab Filter showcasing their Q2 plugin and the "cramping" of bells in typical EQ plugins at lower sample rates, due to the Nyqvist filter cut off, versus the Q2

    This has nothing whatsoever to do with the sampling rate for recording.


    Most plug ins today use internal oversampling, so when you apply the plug in, there is no sound difference whatsoever between material recorded with 44k, 88k, etc.


    Nothing against the (very good) Fab Filter EQ, but their advertising video makes it sound like it's the only EQ on the market that uses internal oversampling, and offers a linear phase option (linear phase is not inherently better, it's better for certain purposes, worse for others..).





    I doubt that anyone can actually distinguish between audio recorded at 44k and 48k. If you think you can, do a double blind test: take any track recorded at 48k, and bounce it, once to a 48k target file, once to a 44k target file. Then do a double blind test, to hear if you can distinguish between the files (and please tell me what you find - I'm genuinely curious..).


    To do a double blind test, you can either use Apple's free AUlab (if you're on a Mac - how-to video here), or any kind of ABX test app.




    Slightly off topic: most DAWs today work at 32bit float internally. In a recent video I watched, Andrew Scheps explained that you can apply a Trim plug in to any DAW track, boost the signal by 80dB, so everything in the DAW's mixing board looks like it's peaking and distorting like crazy. Then you insert another Trim plug in to the sum that reduces the gain by 80dB. Everything will actually sound fine, no distortion whatsoever, due to the way calculations are done internally. Counterintuitive, if you think in analog gain staging terms...

    Quote from stratology

    ........... If you think you can, do a double blind test: take any track recorded at 48k, and bounce it, once to a 48k target file, once to a 44k target file. Then do a double blind test, to hear if you can distinguish between the files (and please tell me what you find - I'm genuinely curious..).....

    That's not what happens in practice though, What you're suggesting addresses sample rate conversion. No one is saying that converting sample rates of the same track will result in audible differences


    It's a different story if you record one at 48k and another at 44.1 and listen to the differences


    If you have a DAW and any drum sampler, start a project and insert a beat and add any reverb plugin then change sample rate from 44.1 to 48k and listen to the drum part. Who cares about theories. Let your ears decide and please report back if you're able as I'm also curious

    Quote from Dean_R

    It's a different story if you record one at 48k and another at 44.1 and listen to the differences.


    If you have a DAW and any drum sampler, start a project and insert a beat and add any reverb plugin then change sample rate from 44.1 to 48k and listen to the drum part. Who cares about theories. Let your ears decide and please report back if you're able as I'm also curious

    No, it's the same.


    If you use a drum sample, that sample was originally sampled at a specific rate, which may or may not be identical with the sampling rate of your DAW project. So it's not really a valid test/comparison, as additional processes take place. Why would a reverb plug in produce different results at 44 or 48k? If you have audio below Nyquist, the audio with reverb added is also below Nyquist.


    So to do a real test, you would have to start with analog material and record completely identical material at 44k and 48k, then do a double blind comparison. Any comparison that is not done in a double blind test is meaningless, due to confirmation bias.



    If you do hear a difference: it's entirely possible that the converter you use works/sounds better at 48k. A different converter may work better at 44k.



    Caring about theories is important, because it lets you determine what is actually going on, e.g. comparing the quality of a reverb plug in (that is already in the digital domain) or a loop (that is also already in the digital domain) vs. comparing the process of sampling of analogue material (A/D conversion) at different sampling rates.



    Once you are in the digital domain, issues with frequencies close to Nyquist are handled by oversampling. If you have a very old plug in that's not capable of that, it's an entirely different issue. Meaning, at the same project sampling rate, a plug in with better design will sound better, and show no preference to any project sampling rate .

    Quote from stratology

    Slightly off topic: most DAWs today work at 32bit float internally. In a recent video I watched, Andrew Scheps explained that you can apply a Trim plug in to any DAW track, boost the signal by 80dB, so everything in the DAW's mixing board looks like it's peaking and distorting like crazy. Then you insert another Trim plug in to the sum that reduces the gain by 80dB. Everything will actually sound fine, no distortion whatsoever, due to the way calculations are done internally. Counterintuitive, if you think in analog gain staging terms...

    As long as you're not using any character plugins (or just slightly badly coded ones) in between =O

    Quote from stratology

    This has nothing whatsoever to do with the sampling rate for recording.


    Most plug ins today use internal oversampling, so when you apply the plug in, there is no sound difference whatsoever between material recorded with 44k, 88k, etc.

    No arguments here. I was merely posting the video that illustrates the "cramping" phenomenon. They do say something along the lines of "typical EQ plugins bundled with your DAW". I've no idea what plugin they used in the video, but some DAWs come with some archaic plugins (Pro Tools 11HD that I use at work, for example). It's one of the reasons why freebie plugins have a reputation of sometimes sounding harsh in the higher registers.

    Quote from Michael_dk

    As long as you're not using any character plugins (or just slightly badly coded ones) in between =O

    Haha, you're right, of course - plug ins that are supposed to introduce (good sounding) distortion will introduce distortion, and badly coded ones will misbehave.



    Sample rate discussions often end in nasty catfights, I'm pleasantly surprised that a good, civilised discussion is possible on the Kemper forum. Great place.



    Just to be clear: I'm fully aware that for someone who works in video, where 48k is standard, being forced to use 44k because the Kemper has to be clock master on SPDIF is disappointing. That's a very good point to make, regardless of possible differences in sound.

    Hi


    I have the Zoom U44 too
    I use it with the Spdif connection
    But I hear a real difference between the sound through monitoring (on Cubase) and the sound on the playback track. far less dynamics on the second one (like if the definition button was turn down drastically)


    Do you know how to solve this problem?


    Thanx

    Hmm interesting thread. Would it be possible, when reamping a recording, to hear the amped sound through the FRFR headphones that is plugged directly in to my soundcard? Since the guitar track is direct and has no effect I figure the DAW's monitor function will only feed a clean signal to my ears when recording?