Harmony Central Forums
Announcement
Collapse
No announcement yet.

Bit rate

Collapse



X
  • Time
  • Show
Clear All
new posts

  • #46
    Originally posted by Rock Austin


    So even if I select the 32 bit recording option, I am still basically recording at 24 bit resolution. I really have no epertise in this realm, but are the VST effects and other processes in the mixing phase done at 32 bit in the computer?

    If true, would using the 32 bit recording option make it easier (or better) for the computer to manage the sound data? Is this the advantage of using this option? (or is there any advantage?)
    If I understand it correctly, Cubase VST 32 is capable of recording 32-bit audio, but all 32 bits are represented correctly if, and only if, your converters (and anything else in your digital audio chain) are also 32-bit. There's still an advantage, however, when it comes to cumulative headroom.

    For example, say we record an acoustic guitar at 16-bit. It sounds good. But then we add a reverb and a delay and an eq. And 48 other tracks with reverb, delay and eq. A ****************ty mix notwithstanding, all of these tracks quickly add up to clipping the stereo mix bus, which'll force you to start turning down channels and aux returns. No problem, that's fine. But at higher bit rates, you're able to turn these signals down and still maintain high sound quality, since the bit rate to begin with was high enough to give you room to move down without worrying about running into all that garbage that lies down in the low levels of a 16-bit signal.

    On a tape recorder, if you record too low, and then turn up the track, you'll increase tape hiss. Likewise on a digital system, if you record too low, and then turn up the track, you'll increase digital garbage. If you record 24-bit (or implement an engine at 32-bit, regardless of what bit rate you're recording at), you're effectively pushing down the volume at which you'll begin to hear that digital garbage. In practice, a smooth fade out at the end of a 16-bit song, if listened to very carefully, will sound a little grainy at the very end, where a 24-bit song will fade out longer before dropping into digital garbage. If you want the reason why, read Duardo's posts. He's definitely knows what he's talking about.

    Comment


    • #47
      I think the basic deal here is that higher bit rates and sampling rates ARE better, and you CAN hear it thru suitable equipment that'll translate to that standard.

      But it costs more, and ur still hearing the same basic material at the end anyway. So until it's cost effective and/or industry standard, there's always gonna be resistance.

      Ima stick with ****************ty 16/44.1 cause im only doing rough demos of my own stuff and some bands, and at this stage for me its more bout getting THE SONG on some sorta repeatable format is the main idea... massively detailed intricate fancy stuff to me is in my head, and if I can't do it with my setup now, then it'll be there when it's being recorded in a schmicko studio n whatnot...
      Forecast for tomorrow; a few sprinkles of genius with a chance of doom!

      I USED to be a señor member...

      Comment


      • #48
        Originally posted by Switch
        I think the basic deal here is that higher bit rates and sampling rates ARE better, and you CAN hear it thru suitable equipment that'll translate to that standard.
        Should you record 24-bit if at all possible? Yes. Should you worry about 96k? No. There are some people on this forum who have insinuated that anything less than 96k is amateurish and pointless if you're doing anything professional. What's funny is that they defend their argument with specs and numbers, NONE of which really reflects real-world sound quality.

        The Beatles used only one mic on their drums, and their records sounded amazing. If anyone told them they needed to record on some crazy, futuristic digital 96k machine to get professional results, they'd laugh in their face.

        But if you want to record at 16-bit, hey, you'll get professional results as well, if you know how to record and mix well.

        Comment


        • #49
          I wouldn't use 96k unless i was recording orchestrated stuff. But ffor Rock i use 24bit/44.1.
          Isn't 88.2 a better number to convert down from since it is exactly twice as much as 44.1?
          NU-TRA! Powered by Logic Audio

          Comment


          • #50
            O.k., time to chime in here.
            I've heard 3 folks in h his thread say all sort of true stuff (The Chinese, Audacityworks and FM),
            and I;ve heard Ethan Whiner and Duardo post some BS, no offense guys.

            Ethan, we CAN sense above 20kHz!, so yes the human HEARING is limited to say 22Hz-18kHz (I've yet to meet anyone over 5 years old, that can point out a pure 20kHz sine wave), but the human SENSE goes probably up to 40kHz.

            And yes, 96kHz HAS a benefit, eventhough you're scaling down to 44.1kHz, and I'll explain why.

            In a 'regular' 44,1kHz system the 'Nyquist freq. is at 22.050Hz. Theoretically there should be a brick wall limiter/filter (with some ridicoulous spec like -1000dB/Oct), but since such filters don't exist and -96dB/Oct filters bring in all sorts of audible artifacts, they've opted for a -48dB/Oct low-pass filter.

            And what does this mean? It means, that if you'd START the filter-roll off at 15kHz, that its -48dB point (of a 96dB scale) is at 30kHz! This is PAST the 22.050 Nyquist freq. And 'aliasing' means that the frequencies PAST the Nyquist freq, a 'mirror' fequency is folded back into the audible range. I can explain why, but it's even more text, and I'm already typing alot...

            This means that between 22.050 and 30kHz, 8kHz of info is folded back (into the 14kHz-22kHz range, which formed the nasty digital harshness in the older systems) .

            With a 96Khz sampling frequency, this means that the Nyquist freq. sist at 48Khz. The low-pass filter is time only a -24dB/Oct one, which is 'smoother'. Say that the same filter starts at 12Khz, this means that the -24dB sits at 24kHz, and the -48dB point sits at 48kHz. So it's only the SOFTEST 48dB of dynamic range that go past 48Khz. If these are folded ack into the 'audible' range, they're most likely MASKED by 'musical info'.

            This means a 20Hz-20kHz with no audible high-end harshness whatsoever, so if this gets digitized and converted back to a CD, you WILL have a clearer representation of the audio information.


            Now, to make full advantage of a pro-96k or even 192k system, you'll need speakers that can produce 15Hz-45Khz. These DO already exist, but imagine the little energy that 40kHz waves have, when being dispersed by a .25" super-tweeter.
            So, yes you'll need better speakers than what you now have

            And yes, 24 bit will give a higher dynamic RESOLUTION, even with a limited a dynamic range. I mean 16 bit gives 65.536 steps of different volume values, 24 bit gives 16million+ values.

            And I'll happily agree with anyone who says that the difference between 16bit/44.1kHz and 24bit/44.kHz is more obvious than the difference between 24bit/44.1kHz and 24bit/96kHz.

            Also, the dynamic 'harshness' is mostly due badly engineered AD converters and anti-aliasingf filters.


            So please, no more "humans can't hear above 20kHz"crap.
            Hell, I tested it myself with an Avalon 737SP pre-amp. That one has a sweepable high-EQ band, that goes up to 32kHz. I let a friend on mine (local post-production engineer), switch the freq. while I was singing/talking into a U87. When he switched it, I could hear/feel more 'air' and breathiness.
            MovingNeedles Recording Studio

            Looking for:
            -MXR Envelope Filter & Blue Box ( both block logo's, +boxes and manuals)

            For sale:
            -Boss SG-1 Slow Gear Manual; VGC, 1979 print. Make me an offer!

            Comment


            • #51
              I've heard Ethan Whiner and Duardo post some BS, no offense guys.


              No offense taken.

              Ethan, we CAN sense above 20kHz!, so yes the human HEARING is limited to say 22Hz-18kHz (I've yet to meet anyone over 5 years old, that can point out a pure 20kHz sine wave), but the human SENSE goes probably up to 40kHz.


              So you're met a five-year-old who can point out a pure 20 kHz sine wave? (Sorry, had to ask.) I'm not aware of any studies that have shown that humans can recognize frequencies about 20 kHz (I know some have shown that the body reacts to them, but not in any conscious way) and since you said that our sense "probably" goes up to 40 kHz I doubt that you have either.

              In a 'regular' 44,1kHz system the 'Nyquist freq. is at 22.050Hz. Theoretically there should be a brick wall limiter/filter (with some ridicoulous spec like -1000dB/Oct), but since such filters don't exist and -96dB/Oct filters bring in all sorts of audible artifacts, they've opted for a -48dB/Oct low-pass filter.


              Who are "they"? That's not the way it's done, at least not these days. Most manufacturers use oversampling converters, which actually sample at a higher rate and thus don't need nearly as steep of an anti-aliasing filter, so all of the phase distortion is well above the eventual Nyquist frequency. Then that oversampled digital filter is run through a digital anti-aliasing filter that is able to roll of frequencies within the 20-22.01 kHz range without any phase distortion.

              And what does this mean? It means, that if you'd START the filter-roll off at 15kHz, that its -48dB point (of a 96dB scale) is at 30kHz! This is PAST the 22.050 Nyquist freq. And 'aliasing' means that the frequencies PAST the Nyquist freq, a 'mirror' fequency is folded back into the audible range. I can explain why, but it's even more text, and I'm already typing alot...


              But since it's not done that way, it doesn't mean anything. If it was done that way then the rolloff would have to start at 5.5125 kHz (in a 44.1kHz system), which obviously isn't the case. You're right, if you started the rolloff at 15kHz using a 48 dB/8va filter you'd only be 48 dB down at 30 kHz and have all kinds of aliasing, which also is obviously not the way it works.

              With a 96Khz sampling frequency, this means that the Nyquist freq. sist at 48Khz. The low-pass filter is time only a -24dB/Oct one, which is 'smoother'. Say that the same filter starts at 12Khz, this means that the -24dB sits at 24kHz, and the -48dB point sits at 48kHz. So it's only the SOFTEST 48dB of dynamic range that go past 48Khz. If these are folded ack into the 'audible' range, they're most likely MASKED by 'musical info'. [quote]

              No, this isn't the way it works at all. You can't start rolling frequencies off at 12 kHz and expect that converter to be accepted on anything close to a professional level. That would mean it would be about 12 dB down at 18 kHz, which would sound to us as if the volume were approximately cut in half. And even if the aliased frequencies are well below the average level, they tend to be unharmonic and are therefore still likely to be perceived. But your reasoning does make sense, and that's why oversampling filters are used...you can begin to roll frequencies off at 20 kHz and do it slowly and gently over many octaves, and then do your digital filtering once the signal's been converted with a much steeper slope.

              [quote]This means a 20Hz-20kHz with no audible high-end harshness whatsoever, so if this gets digitized and converted back to a CD, you WILL have a clearer representation of the audio information.


              In your incorrect example, if all frequencies were rolled off at -24 dB/8va starting at 12 kHz, you would most certainly not have a clear representation of the audi oinformation.

              Now, to make full advantage of a pro-96k or even 192k system, you'll need speakers that can produce 15Hz-45Khz.



              Another issue entirely.

              And yes, 24 bit will give a higher dynamic RESOLUTION, even with a limited a dynamic range. I mean 16 bit gives 65.536 steps of different volume values, 24 bit gives 16million+ values.


              Again, that makes sense intuitively...but when you consider that ALL of that extra resolutio is below -96dBFS, it's ONLY an issue if the dynamic range of your material exceeds 96 dB (theoretically...in actual practice a bit less, but not much).

              l happily agree with anyone who says that the difference between 16bit/44.1kHz and 24bit/44.kHz is more obvious than the difference between 24bit/44.1kHz and 24bit/96kHz.


              I will too, but only in those cases where the material has an extremely wide dynamic range. Your average pop/rock recording? No way.

              he dynamic 'harshness' is mostly due badly engineered AD converters and anti-aliasingf filters.


              What is "dynamic harshness"?

              So please, no more "humans can't hear above 20kHz"crap.
              Hell, I tested it myself with an Avalon 737SP pre-amp. That one has a sweepable high-EQ band, that goes up to 32kHz. I let a friend on mine (local post-production engineer), switch the freq. while I was singing/talking into a U87. When he switched it, I could hear/feel more 'air' and breathiness.


              You weren't hearing anything up at or near 32 kHz, what you were hearing was the effects of the filter at and below 20 kHz (or maybe 18, or 16, or whatever the upper limit of your hearing is). You can prove this to yourself by repeating your test and recording it at 44.1kHz. You'll hear that "air" and "breathiness" even though you now know there's no information above 20 kHz at all.

              -Duardo

              Comment


              • #52
                It took me some minutes to stop laughing after I read your posts.


                I'm glad I was able to provide some amusement for you. Can't say you haven't done the same for me.

                Every audiowave that goes up and down, and crosses the zeropont, is really loud, and gets really really quiet when it goes near the zeropoint before it crosses it.
                So your point is BS.


                That's how it's represented, sure. It crosses the zero crossing, but never stays there.

                Do you know what "dynamic range" means? To put it simply, if we're talking about a recorded signal, it's the difference in level between the loudest and quietest parts of that signal. If I have a keyboard and hold a note down that sustains at the same level until I release the key, we would say that the signal has a dynamic range of zero, even though if we looked at a representation of that signal as a waveform, it would cross over the zero crossing multiple times. It doesn't "get quiet". Are we at least agreed there?

                In Sampledata we count UP and if signed we ALSO count down.
                The dBFS scala is just an INTERPRETATION, so that you can have digital VU- and Peak meters for example.


                No, we only count down. As I said in my last post, although we look at a wave on a graph that oscillates over and under a zero crossing, the way we represent this is in dB below FS. You can't argue that. Did you try what I recommended and run a 0 dBFS signal out of a 16-bit machine into a 24-bit machine? It doesn't show up at -48dBFS on the meters, does it? No. 0 dBFS=0dBFS no matter what the bit rate.

                There is NO SINGLE DSP process that cares about dBs. SampleData has NOTHING GOT TO DO with Dbs.
                The SampleValues are only DISPLAYED in dBFS, cause we are used to it.


                Sure, the DSP process itself is just math. In this case, though we do express things in dB of some sort because that's what we have to work with. Until our signal hits a D/A converter we have to express it in dBFS. How else could we express it?

                Oh....that are news. The official definition of 0dB SPL is there is NO MOVEMENT OF AIR AT ALL leading to ZERO PRESSURE.
                What has that got to do with the average human ???


                No, you are absolutely wrong. Are you making up your own definitions? 0 dB SPL is equivalent to a sound pressure of 0.00002 per square centimeter. This is equivalent to 0.000000000001 watts per sqare meter. 0 dB SPL is indeed based on the hearing capability of the average human. What good would it do to reference sound in a vacuum? None of us spend any time there.

                -> OMG! As you should have seen, I added the additional digit to the RIGHT, not as you did to the left...hehe
                If it would get added like you said to the "left", then a 16 bit 0dBFS data in a 24 bit dataslot would equal approx -48dBFS like I already said.


                So what are you saying here, that you were wrong then or that you're wrong now?


                --------------------------------------------------------------------------------
                we're capturing those signals closer to the zero crossing with more resolution.
                With 16 bits we can theoretically capture signals up to 96 dB below FS accurately
                --------------------------------------------------------------------------------

                -> Absolute Bull****************. Really. A very common mistake.


                No, this is still right.

                Btw, it are 6,02 dB per bit, not just 6dB/bit. Just for the record...


                Yes, I know this. Actually, if you want to get technical about this, the forumla to figure out the possible dynamic range in dB of a system with a dynamic range of x is:

                6.02 (x) + 1.76

                So a 16-bit system theoretically has a 97.98 dB dynamic range, although that's a theoretical and not real-world example, so it doesn't relaly matter for the purpose of this discussion...and for our discussion here, 6 dB/bit is close enough. Agreed?

                In the following example We translate FULL SIGNAL to dBFS,
                and then we just go ONE VALUE (not one bit) below Full signal,
                and finally the smallest value that is not zero gets translated.

                (Duardo here...I cut out the code...scroll down if you want to see it...)

                The first thing you should realize, is easy to see, when you compare the red and the green value.
                Going just ONE little unit below the MAXIMAL VOLUME representation is 256-times MORE EXACT when happening in 24 bit (green) than in the 16 bit (red) system.
                (If you dont see it, then divide the red value by the green value...Voil

                Comment


                • #53
                  make it stop....please make it stop....my head is hurting....rage....rising.....please.......make... ...it........stop..........
                  "That's a matter of opinion. Like "a supermodel is less attractive than a sore-infested, poo-covered morbidly obese bald chick" is a matter of opinion."
                  - Audacity Works

                  "Ain't nobody ever walked down the street humming the sound of a microphone... it's all about 'the music'."
                  -Fletcher

                  Comment


                  • #54
                    If you want to make it stop hurting, look at the "Bit Rate Part II" thread...I think I've explained myself there in a way that will be much easier to comprehend...

                    -Duardo

                    Comment


                    • #55
                      SD,


                      As Duardo explained, that is not how A/D conversion is done these days.


                      And the speakers used by the consumers / listeners?

                      I loved the comment elsewhere that Sergeant Pepper's was recorded using gear that by today's 44/16 standards is woefully inadequate - yet it still sounds great!


                      Let me explain the basics of filter Q, or bandwidth. Regardless of a filter's center frequency, other adjacent frequencies are also affected. So if you apply a 32 KHz. filter and the Q is not too high, frequencies within the audible band are also affected. That is what you heard.

                      --Ethan
                      The acoustic treatment experts
                      Buy my DVD

                      Comment


                      • #56
                        Call me crazy, but those Pizza Hut bread sticks are maybe the world's most perfect food !

                        Comment


                        • #57
                          Originally posted by Ethan Winer
                          SD,


                          As Duardo explained, that is not how A/D conversion is done these days.

                          Can you or anyone else explain how it IS done these days? From my understanding, -96dB/Oct filters bring in audible artifacts, so that's a reason not to use those.

                          Originally posted by Ethan Winer

                          And the speakers used by the consumers / listeners?

                          Point taken, BUT... We all act like audiophiles and try to make the best sounding mixes possible, so that the 0.0000001% of the consumers (the ones with really good equipment) can fully enjoy our music. I mean, if we would think "Oh, Joe Doe is just gonna listen to my CD on a crappy PC-set up or mp3's", we might as well alltogether drop the $20k+ ProTools HD gear and get back to our beloved Tascam Portastudio's...
                          Get MY point?
                          sounds great!

                          Originally posted by Ethan Winer

                          Let me explain the basics of filter Q, or bandwidth. Regardless of a filter's center frequency, other adjacent frequencies are also affected. So if you apply a 32 KHz. filter and the Q is not too high, frequencies within the audible band are also affected. That is what you heard.

                          --Ethan

                          Agreed, but I think there's more to it. BTW, his WHOLE signalchain was pretty damn high-end. U87 into 737SP into Apogee Trak2 (or whatever that red converter is called) into ProTools TDM into Genelec 1031A's...
                          MovingNeedles Recording Studio

                          Looking for:
                          -MXR Envelope Filter & Blue Box ( both block logo's, +boxes and manuals)

                          For sale:
                          -Boss SG-1 Slow Gear Manual; VGC, 1979 print. Make me an offer!

                          Comment


                          • #58
                            Can you or anyone else explain how it IS done these days? From my understanding, -96dB/Oct filters bring in audible artifacts, so that's a reason not to use those.


                            Do you actually read my answers to your post? I'll answer briefly here, but check earlier posts for more detail.

                            First, the analog signal is passed through a filter with a very gentle rolloff starting at about 20 kHz. The filtered audio is then sampled at a rate that's a multiple of the actual sampling rate being used...maybe 4x, maybe 8x, maybe more...this raises the Nyquist frequency enough that any phase distortion artifacts will be well beyond the audible range. So now we have a bandlimited signal that rolls off completely at, say 176.4 kHz in the case of an 8x oversampled 44.1kHz converter (8x44.1=352.8 kHz, half of which is 176.4). Now that this is a digital signal, we can filter it again, but this time we can use a digital filter that isn't subject to the limitations of analog filters...so we can start it at 20 kHz and remove everything below 22 kHz without audible artifacts. Now we have a signal at a 352.8 kHz sampling rate that has no information above 22 kHz. We then downsample, or decimate, this signal so it's a 44.1 kHz signal. Since there's no information above 22 kHz, all we have to do is drop seven out of every eight samples and we still have the exact same waveform. Exactly.

                            That is how it's done. Ethan, please correct me if I'm wrong.

                            Agreed, but I think there's more to it.


                            You're making an assumption there. Many people make that same assumption. Several very well-respected analog designers have made statements to that effect, that their gear needs to be flat well beyond 20 kHz to sound right to us. And that's true. The error they make is that that applies to digital conversion as well. As I said, it's very easy to test this. Take your chain you're talking about, run it into the Trak2 at 44.1kHz, hit record, tweak that 32 kHz knob, and then listen back to the recording. You'll hear it. It's that simple.

                            -Duardo

                            Comment


                            • #59
                              Speed,


                              I think Duardo did - a few times!


                              And this is exactly the problem! "From my understanding" implies you are not speaking from direct experience, but merely parroting what you've heard others say. Maybe even famous engineers you read interviewed in magazines. Even more to the point, what audible artifacts? What do those artifacts sound like, and how do they negatively affect audio quality?


                              Agreed. But at some point using ever-higher bit or sample rates ceases to offer any audible improvement. Past that it's just a big waste of hard disk space and computing power. Now that we both agree that eventually a further increase in resolution is pointless, we can now get on to determining where the cutoff point it. I suggest you test some of this stuff for yourself and stop believing everything you read and hear.

                              EW: So if you apply a 32 KHz. filter and the Q is not too high, frequencies within the audible band are also affected. That is what you heard. <<


                              What more to it? If you don't have a shred of evidence, or any direct experience, or even a hint of an explanation, then what point are you arguing and why?


                              That's totally irrelevant. Anyone can buy a bunch of expensive gear. That doesn't make them knowledgeable.

                              --Ethan
                              The acoustic treatment experts
                              Buy my DVD

                              Comment


                              • #60
                                Duardo,


                                Ha ha. You're doing fine all by yourself!

                                --Ethan
                                The acoustic treatment experts
                                Buy my DVD

                                Comment













                                • Working...
                                  X