Samplerate discussion

This discussion was created from comments split from: Buffer bug.

Comments

  • It’s better not to use 96khz in iPad realm, cpu is not strong enough and anyways, it doesn’t add anything except for lower latency.
    48khz is more than enough for post production even though BM3 workflow and sequencer is not meant for it.
    But I definitely agree that the option should be available as new iPads will soon emerge.
    I’d be better off with buffer going down to 64 just for live monitoring audio with fx.

  • edited October 19

    @bargale said:
    It’s better not to use 96khz in iPad realm, cpu is not strong enough and anyways, it doesn’t add anything except for lower latency.
    48khz is more than enough for post production even though BM3 workflow and sequencer is not meant for it.
    But I definitely agree that the option should be available as new iPads will soon emerge.
    I’d be better off with buffer going down to 64 just for live monitoring audio with fx.

    This doesn't hold true for me, i make sample packs, they have to be 96khz for a lot of reasons, this makes B3 unusable for creating sample content, i need 96khz.
    Also saying it doesn't add anything isn't true, depending on how any particular DSP is coded (For instance AU plugins) 96khz will be better quality, modulations will be smoother etc.

  • @5pinlink said:

    @bargale said:
    It’s better not to use 96khz in iPad realm, cpu is not strong enough and anyways, it doesn’t add anything except for lower latency.
    48khz is more than enough for post production even though BM3 workflow and sequencer is not meant for it.
    But I definitely agree that the option should be available as new iPads will soon emerge.
    I’d be better off with buffer going down to 64 just for live monitoring audio with fx.

    This doesn't hold true for me, i make sample packs, they have to be 96khz for a lot of reasons, this makes B3 unusable for creating sample content, i need 96khz.
    Also saying it doesn't add anything isn't true, depending on how any particular DSP is coded (For instance AU plugins) 96khz will be better quality, modulations will be smoother etc.

    Modulation sample rate (ie-LFOs) is separate from audio-signal sample rate.

    Or are you talking about DAW automation? If so, please explain/educate.

    On a digital synth, the audio signal path and modulation paths usually run at different sample rates. For example, the Nord G2 uses 96kHz to process audio, whereas low-frequency control signals & modulations (ie-LFOs) run at 24kHz. This is because they are often low-frequency signals by nature, and do not require a high bandwidth. All signals (audio, control, logic) run at 24bit in the Nord G2, for example.

    On other synths, some parts/modules in the audio signal-path may be over-sampled (ie-filters) but are down-sampled at the 'output' to 96kHz or less. Many digital reverbs do this (2C Aether) and sound just average with over-sampling turned off, but sound significantly more detailed with it enabled.

    Then there are the folks who say that a plugin should be coded to sound the best with no internal over-sampling 'hacks'.

    Which iOS apps were coded to work best at 96kHz?

  • According to Bram bos everything is done at sample rate in AU for instance, that is just the way IOS works, not sure of the relevance though, my point was that saying 96khz doesn’t add anything, is simply untrue.

    My assertion that i personally need 96khz actually has nothing to do with IOS, and more to do with creating samples.

  • That is interesting. If all is done at sample rate in AUs on iOS, then sample rate is significant. Seems like a waste of CPU cycles though. You could route the non-audio signals in the Nord G2 to audio modules, and the non-audio signals would then run at 96kHz, but the CPU usage would jump up a lot, and there wouldn't be any different when measuring on an oscilloscope.

    Anyway, yup, I understand when you say 96kHz is your preference when sampling. I like to record fingerstyle acoustic guitarists at the highest possible resolution. Big FX-drenched soundscapes & metallic atmospheres too. I played around with the DSD 1bit 5.6mHz stuff too, but convertng to PCM to edit it made it not worth it.

    OK, back to topic (Buffer settings):
    With an empty BM3 Session:
    2013 iPad Mini 2, I try setting it to 64 samples, it switches to 128.
    2011 iPad 2, lowest is 256.

  • @5pinlink said:

    @bargale said:
    It’s better not to use 96khz in iPad realm, cpu is not strong enough and anyways, it doesn’t add anything except for lower latency.
    48khz is more than enough for post production even though BM3 workflow and sequencer is not meant for it.
    But I definitely agree that the option should be available as new iPads will soon emerge.
    I’d be better off with buffer going down to 64 just for live monitoring audio with fx.

    This doesn't hold true for me, i make sample packs, they have to be 96khz for a lot of reasons, this makes B3 unusable for creating sample content, i need 96khz.
    Also saying it doesn't add anything isn't true, depending on how any particular DSP is coded (For instance AU plugins) 96khz will be better quality, modulations will be smoother etc.

    Nyquist Theorem - https://whatis.techtarget.com/definition/Nyquist-Theorem
    If you really want higher Sample rate for audio usage it is more logical to go for 88.2 KHz as it it twice the rate of the audio industry standard - 44.1 KHz.

    P.s
    If you’re working only in the digital realm than there is no conversion process.
    The only recommendation as for working non standard is by going to 32bit floating point audio as there’s no chance of digital distortion in case of bad audio processing.

    P.s 2
    I actually work in 48khz in bm3 because of the buffer bug.

  • edited October 20

    44.1 has not been any standard for a long time, that was back in the days of CD.
    48 96 and 192 are actually closer to a universal standard now because of DVD/Bluray/TV/Consoles which are a lot more common in households than a CD player.

    So at this point any division of 192 is actually the future proofed variant, 24/96 being the accepted current middle common ground for recording, however, that being said, a few mastering houses have already started asking for 32/192 masters depending upon what the final format will be, a lot of game houses are now requesting 32/192 masters as are post pro houses too.

  • edited October 20

    As for 32 bit audio floating point being a recommendation for no distortion, again that is not true either, up until fairly recently the studio standard (Pro tools) was Integer and not floating point audio engine, a lot of engineers will still only use integer too, floating point can never truly be distortion free because it will always have a fixed number of decimal places, this means that some rounding ALWAYS has to take place in any calculation, honestly arguments can go on for days in either camp, i have been part of them over the years.

    All these points are in actual fact complete and utter BS, why, because we neither record or mix in a perfect acoustic chamber that warrant anybody worrying about any of this rubbish.

    Heres the rules, client wants 24/96, so i NEED 24/96 lol

  • I split this off, didn't want to drown the original bug thread, which is a valid bug ;)

  • What's funny is that several of the highly regarded audiophile DACs aren't really 192kHz or 384kHz, they internally upsample via software to get those numbers. Plus I've talked to several canines that critically listened to them, and they told me they couldn't hear a difference.

    You hear about better 'air', detail, instrument seperation, soundstage width & depth, etc. especially in the high frequencies with ultra-high sample rates, especially with the Korg 1bit DSD recorders at 5.6mHz, but what's rarely mentioned is that the best mics and preamps were used, plus the meticulous acoustic treatment of the physical spaces.

    Also let's not forget that our ears & brains are accustomed or 'tuned' over time to expect certain things, so some music genres will sound 'too good' when recorded at ultra-high resolutions, like harder rock music. IMO, it's probably best suited for recordings of live classical & jazz music.

    Though I have some 24bit 192kHz remasters of The Beatles, & I heard new things I haven't heard before like people talking in the background of the studio, John Lennon giggling all the time, especially when George Harrison rips into a solo, Ringo's many accidental hits, cursing under their breath, etc.

    I used to be obsessed with sound quality, and usually the only outcome is massive financial debt, IMO. That extra 2% of quality you get when paying 1000% more is usually just your brain justifying your purchase. It's a vicious cycle for some folks. It's pretty insane.

  • @TONBOGIRI said:
    What's funny is that several of the highly regarded audiophile DACs aren't really 192kHz or 384kHz, they internally upsample via software to get those numbers. Plus I've talked to several canines that critically listened to them, and they told me they couldn't hear a difference.

    You hear about better 'air', detail, instrument seperation, soundstage width & depth, etc. especially in the high frequencies with ultra-high sample rates, especially with the Korg 1bit DSD recorders at 5.6mHz, but what's rarely mentioned is that the best mics and preamps were used, plus the meticulous acoustic treatment of the physical spaces.

    Also let's not forget that our ears & brains are accustomed or 'tuned' over time to expect certain things, so some music genres will sound 'too good' when recorded at ultra-high resolutions, like harder rock music. IMO, it's probably best suited for recordings of live classical & jazz music.

    Though I have some 24bit 192kHz remasters of The Beatles, & I heard new things I haven't heard before like people talking in the background of the studio, John Lennon giggling all the time, especially when George Harrison rips into a solo, Ringo's many accidental hits, cursing under their breath, etc.

    I used to be obsessed with sound quality, and usually the only outcome is massive financial debt, IMO. That extra 2% of quality you get when paying 1000% more is usually just your brain justifying your purchase. It's a vicious cycle for some folks. It's pretty insane.

    I, too, was obsessed about sound quality but nowadays a 100$ interface will be better sounding than the top of the line equipment being used in the 60’s-90’s era.
    This is the reason why I moved from the Mac to the ipad,
    there’s really no need for more than that to make great sounding music.

  • edited November 4

    It's true, on iOS everything can run at samplerate. It's up to the developer to make wise decisions about it.

    E.g. I typically oversample frequency-critical stuff like filters, FM and oscillators (running 96KHz at 4x oversample rate will certainly heat up your CPU :D ) and undersample modulations I know are going to be slow. You won't hear stepping for a really slow LFO if it updates at 12KHz instead of 48KHz... :)

    Typically in my own experience it's more useful to run at higher samplefrequencies than to run at higher bitrates. Those last few bits of dynamic range are usually so far below the noisefloor that only superheroes from Krypton will be able to hear them - especially when you're just recording samples (when mixing 60 layers of orchestral recordings it's a different matter, obviously :D ).

  • @brambos make sure you visit more often mate, there has been some limited panic in the chat about you not showing up (I believe the current theory was, invented a time machine but it only worked one way and then broke down)

  • edited November 4

    Welcome back 'Bos man'.

    We certainly missed you here

  • There was a lot of samplerate discussion with regard to the channel saturation in the Harrison Mixbus DAW (I’m using Mixbus on Linux alongside BM3 myself, so keep up with it’s developments).

    I got a lot of insight into the practicalities of samplerate in terms of dealing with aliasing from these two articles.

    Firstly can see a description with video here showing this in effect:

    http://www.admiralbumblebee.com/music/2017/09/18/Mixbus-32c-The-Mixer.html#tape-saturator

    Looks pretty bad!

    I later found this thread on the Harrison message board to be very insightful in terms of putting this info into practice in a real life context:

    http://mixbus.harrisonconsoles.com/forum/thread-5605.html

    From these two sources, I took a pretty clear knowledge of an actual potential effect of samplerate decision. I think that this reply there sums it up well:

    96kHz is good. 44.1kHz is good enough. So no need for 96k and the extra space and I/O and DSP power 96k needs.

    Spend your money rather on the analogue ends - in-chain and room treatment/monitoring.

    Personally, with this in mind I just roll back the channel saturation a bit on busses with a lot of high frequency content and don’t spare it any more thought.

  • @5pinlink said:
    @brambos make sure you visit more often mate, there has been some limited panic in the chat about you not showing up (I believe the current theory was, invented a time machine but it only worked one way and then broke down)

    I'm still here! Just really busy with life and stuff. I'll make sure to stop by more often :)

  • @brambos said:

    @5pinlink said:
    @brambos make sure you visit more often mate, there has been some limited panic in the chat about you not showing up (I believe the current theory was, invented a time machine but it only worked one way and then broke down)

    I'm still here! Just really busy with life and stuff. I'll make sure to stop by more often :)

    Noir is all across Facebook/Discord, trying to put in to words how off the chain it is is hard work lol, cant wait to buy it, cant wait for these guys to buy it, Really feels like a step up in AU development to me !!!

  • We're all queuing up to instabuy this one @brambos

    Make sure to set a price that rewards your efforts appropriately.

    <3

  • edited November 5

    Noir seems to be a game changer, once again, @brambos !

  • Hah.. can't see Facebook anymore since I left (anything weird happened?), but I just joined the Discord server so I'll be over there too from time to time.

  • Don't worry, i'm on FB spit to help out on the Beatmaker page there, i'm already cluster bombing it with stories of the mythical Noir !!

  • Awesome! B) >:) :D

Sign In or Register to comment.