Harmony Central Forums
Announcement
Collapse
No announcement yet.

Bit rate

Collapse



X
  • Time
  • Show
Clear All
new posts

  • #31
    Todd,


    How, through our armpits? That simply is not true. And it's amazingly simple to test this. Have you ever tested this? Please explain your testing method and equipment setup.


    Again, this is simply not true. Errors in conversion between analog and digital show up as distortion. Plain old 16 bit 44.1 KHz. has extremely low distortion. Far lower than anything you or I or anyone else can hear.

    --Ethan
    The acoustic treatment experts
    Buy my DVD

    Comment


    • #32
      Originally posted by Duardo


      Why does that sound unlikely? It's a very simple process mathematically...you basically just have to drop every other sample and you're there. Converting to 44.1 from 96 is much more complex mathematically, and that's why there are plainly audible artifacts on cheaper converters or poorly-written code.

      -Duardo
      Yep. And for those of you interested, the reason the first DAT machines were only 48k was that at the time, it was extremely difficult to write SRC algorithms which could convert to 44.1, so people couldn't copy CD's. 48,000 was mathematically nowhere close to being divisible (in any iteration) with 44,100.

      At least that's the myth.

      Comment


      • #33
        Originally posted by Duardo


        The thing is, with today's good converters there really isn't the need to get a hotter signal for a hotter signal's sake. Again, using the example of a signal with a 70 dB dynamic range (which is very wide), with a good 24-bit converter it doesn't matter if you're peaking at -1 dBFS, -10 dBFS, or -20 dBFS...you're capturing that signal with the exact same precision. Not that there aren't other reasons to do so...especially if you're mixing through an analog board...but not to "maximize the bits" or "get more resolution" or "use more steps".



        Deeper bit depth does not capture it better unless the dynamic range is so wide that it exceed the limitations of the 16-bit system. If the 16-bit system can capture it accurately, the 24-bit system won't capture it "better". And the self noise of the system won't be an issue unless the dynamic range of the signal approaches the limits of the signal. With that same theoretical signal with a dynamic range of 70 dB, if you record it peaking at, say, -2dBFS, you're capturing with all the accuracy you need and the self-noise of the system is still low enough you won't hear it. If you record it at 24 bits you're not gaining anything, except maybe the ability to set your levels a little more conservatively and peak at -10 dBFS or lower so you don't have to worry about accidental overs or anything like that.

        -Duardo

        Yes-thats ectly what I am referring to-Being able to lower your record level without sacrificing quality. Een in music, since I started using a good 24bit converter, I haven't felt the need for up to the max levels. Instead I record at 0VU.
        My Rig:

        EAW LA325's (2)
        EAW LA400's (4)
        QSC PLX 3402 (2)
        QSC PL1.8
        QSC RMX 2450
        Ashly ProTea 3.24c
        Ashly ProTea 4.24G
        EAW LA212 Monitors
        JBL PRX-512M Monitors
        Whole Lotta Cables

        Comment


        • #34
          Originally posted by Ethan Winer
          Todd,


          Again, this is simply not true. Errors in conversion between analog and digital show up as distortion. Plain old 16 bit 44.1 KHz. has extremely low distortion. Far lower than anything you or I or anyone else can hear.

          --Ethan


          I never said that there were errors in conversion, I said that it is more accurate. There's a difference. If you are happy and content to record at 44/16, be my guest. There is a quantifiable difference betwenn 24bit and 16 bit audio. If you can't hear the difference, then I would suggest that your monitoring needs attention or converters are bad. If you still can't hear the difference, then I suggest a hearing check/new career.
          -Todd A.
          My Rig:

          EAW LA325's (2)
          EAW LA400's (4)
          QSC PLX 3402 (2)
          QSC PL1.8
          QSC RMX 2450
          Ashly ProTea 3.24c
          Ashly ProTea 4.24G
          EAW LA212 Monitors
          JBL PRX-512M Monitors
          Whole Lotta Cables

          Comment


          • #35
            Originally posted by audacity works

            I definitely believe 24-bit recording will give almost everyone a noticable boost in sound quality (and it is here where the connection to headroom lies, and not in sample rate), but if, with professional gear, we can't hear but the tiniest bit of difference between 44.1 and 96k, it's definitely my place to defend the easily corruptable newbies when someone tells them 44.1 has no place in professional audio. Hundreds of world-class facilities would strongly disagree.

            But for film/post and DVD authoring, I guess I could see where 96k is appropriate. I mean, you guys have million-dollar budgets!


            Actually, The Film/Post world will be the last world to convert over to 88/96 recording. Too much infrastructure is in place to change this quickly. And the simple truth is that 48k is perfectly fine w/ most clients (except the Larry blake's, which is cool, and why we need to be ready).
            Million dollar budgets my ass.
            Maybe on paper, but not in reality.
            -Todd A.
            My Rig:

            EAW LA325's (2)
            EAW LA400's (4)
            QSC PLX 3402 (2)
            QSC PL1.8
            QSC RMX 2450
            Ashly ProTea 3.24c
            Ashly ProTea 4.24G
            EAW LA212 Monitors
            JBL PRX-512M Monitors
            Whole Lotta Cables

            Comment


            • #36
              Todd,


              And the difference is ... ?


              Sure, if your record meters never get much above -30 dB.


              I assure you there's nothing wrong with my monitors or my hearing.

              I will concede that using 24 bits makes sense with some sources, such as when recording chamber music or an orchestra. But even then, assuming sensible recording levels, the room noise is more likely to be the limiting factor.

              --Ethan
              The acoustic treatment experts
              Buy my DVD

              Comment


              • #37
                Originally posted by The Chinese
                Yes-thats ectly what I am referring to-Being able to lower your record level without sacrificing quality. Een in music, since I started using a good 24bit converter, I haven't felt the need for up to the max levels. Instead I record at 0VU.


                [sarcasm] No no no! Use a limiter to track everything as close to full scale as you possibly can! Otherwise, you are throwing away bits, and you wouldn't want to waste any, would you? WOULD YOU?! [/sarcasm]

                I track at 0, too. I started by accident--I thought for sure I ruined a take by not getting close enough to full scale once, but it turned out to be one of the best sounding tracks I had ever recorded. My pres thanked me, my outboard gear thanked me, my plug-ins thanked me, and my master fader stayed parked at zero. Yes, my gear talks a lot.

                Comment


                • #38
                  Originally posted by Ethan Winer
                  Todd,


                  And the difference is ... ?


                  Sure, if your record meters never get much above -30 dB.


                  I assure you there's nothing wrong with my monitors or my hearing.

                  I will concede that using 24 bits makes sense with some sources, such as when recording chamber music or an orchestra. But even then, assuming sensible recording levels, the room noise is more likely to be the limiting factor.

                  --Ethan


                  This is my last post on this-
                  A)-The difference is that one is a mistake, whereas the other is a guess, based on where the waveform is.

                  B)-There is a quantifiable difference, period. recording at 24bit allows me to record at 0VU with better results

                  C)-If not your hearing, if not your monitors, then everyone else in recording is wrong. A guy who makes 100k in the biz couldn't be right, could he?

                  -Todd A.
                  My Rig:

                  EAW LA325's (2)
                  EAW LA400's (4)
                  QSC PLX 3402 (2)
                  QSC PL1.8
                  QSC RMX 2450
                  Ashly ProTea 3.24c
                  Ashly ProTea 4.24G
                  EAW LA212 Monitors
                  JBL PRX-512M Monitors
                  Whole Lotta Cables

                  Comment


                  • #39
                    Yes-thats ectly what I am referring to-Being able to lower your record level without sacrificing quality. Een in music, since I started using a good 24bit converter, I haven't felt the need for up to the max levels. Instead I record at 0VU


                    Exactly. Most people don't realize this.

                    There is a quantifiable difference betwenn 24bit and 16 bit audio. If you can't hear the difference, then I would suggest that your monitoring needs attention or converters are bad. If you still can't hear the difference, then I suggest a hearing check/new career.


                    That statement needs to be clarified a little...there is a difference, but only when your dynamic range gets real wide. You said as much yourself...you like 24 bits because you can lower your record level without sacrificing quality. That's the same thing as recording at a lower bit rate (in fact, that's exactly what it is). And the fact of the matter is, for probably 99 per cent of the music recorded these days, 16 bits is more than sufficient and there is no noticeable increase in quality or resolution gained by recording at 24 bits (although there are plenty of reasons to do so as we've mentioned several times).

                    There may be a quantifiable difference for you, but there's not for everybody.

                    If not your hearing, if not your monitors, then everyone else in recording is wrong. A guy who makes 100k in the biz couldn't be right, could he?


                    There are plenty of people making a lot more than that in "the biz" who certainly make mistakes. And believe it or not, not "everyone else in recording" agrees with you. And I bet most people who do haven't even done any sort of listening tests themselves...they just figure that 24 bits must be better because that's what they've read. And there are plenty of valid reasons for recording at 24-bit resolution (I typically do so myself) but most of the reasons people cite most often...from the high-school kid who just picked up his first issue of EQ to the "big guys" in the industry who are interviewed in that magazine...are not correct.

                    -Duardo

                    Comment


                    • #40
                      I use a Delta 1010, which is 24bit/96KHZ recorded into Cubase 5.1/VST32. I assume that the 32 bit record option in Cubase isn't a true 32bit, since the Delta is 24bit. Is this "faux" 32bit?

                      (This discussion reminds me of the good 'ole days when Geordi and Wesley Crusher would go at it for an entire episode....I was lost after about 3 posts..
                      "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


                      • #41
                        FM (why won't you tell us your name?),


                        Yes, and the loss of precision shows up as distortion. But even using only 16 bits the distortion is 10-100 times less than any loudspeakers!

                        --Ethan
                        The acoustic treatment experts
                        Buy my DVD

                        Comment


                        • #42
                          For example when you would have a fully used 16-bit signal, and insert it into 24-bit dataslot, you would increase your socalled Headroom of 48dB (8*6dB). Which means the upper 8 bits of the 24 bitslot are empty.


                          No, you've got it backwards. You get those extra 48 dB (theoretically) below the 96 dB you've already got, not above. It's very simple to test this. Take a 16-bit signal that peaks at 9 dBFS (your "fully used" 16-bit signal) and run it into a device that has a 24-bit digital input. Does it peak at -48dB? No, it peaks at 0. 0 dBFS is always 0dBFS, whether it's on an 8-bit signal or a 24-bit signal. You can't go any higher, just lower.

                          BUT a real 24 bit signal doesnt say anything about its REALWORLD dynamic range. I mean you cannot say that a full 24 bit signal means automatically 144dB SPL.


                          No, of course you can't...but if you had a system that was capable of reproducing 144 dB SPL and played a 24-bit signal at full scale, then played a 16-bit signal at full scale, it'd be just as loud. But if you had a signal that got quieter and quieter, the 16-bit signal would disappear into noise much sooner than the 24-bit signal would.

                          Now we use a ruler with a cm-scale. It shows 100 cm. The smalest unit on it is 1cm. This should be the i.e. 16-bit ruler.

                          Another ruler has a mm-scale. It shows 1000 mm. The smalest unit on it is 1mm. This is i.e. the 24-bit ruler.

                          Now it is easy to see. The cable is still the same, and still is 1m.
                          But the mm-scale has a smaller resolution and is therefor MORE exact than the cm-scaled ruler.


                          That analogy is simply wrong. It makes sense intuitively, but it's not right. I'll give you a grossly oversimplified analogy that should nonetheless make sense. Say you want to express a whole number. It takes two digits to express any number between zero and ninety-nine. It takes three digits to express any number between zero and nine hundred ninety-nine. But using three digits to express a number below one hundred is no more accurate than doing so with two, just like using 24 bits to describe a signal with a dynamic range of up to 96 dB is no more accurate than using just sixteen (theoretically, of course...due to quantization noise and other factors you can't use the full 96 or 144 dB, but we're talking real-world here and are talking about the resolution within that dynamic range).

                          Would one of you say, "No, we dont need a 1000-unit ruler, 100 units are enough"


                          If your analogy were correct, sure, but it's not. First off, in your analogy both rulers are the same length. If the 16-bit ruler measured 100 cm (or 1000 mm) the second one should measure 150 cm (or 1500 mm). The one thing we can all agree on is that the theoretical maximum dynamic range of a 16-bit recording is 96 dB and that the theoretical maximum dymanic range of a 24-bit recording is 144 dB, so the 24-bit ruler should be 1.5 times the length of the 16-bit ruler.

                          Second, in neither ruler would the gradations be evenly spaced. Your analogy is a linear example, and sound (as expressed in dB) is logarithmic. As we went from the top of the rulers down to the bottom, the gradations would get closer and closer together. They would line up perfectly until we got to the bottom of the shorter ruler, after which they would continue to get closer and closer together until they got to the bottom of the second ruler. In this example, the 16-bit ruler is divided into 65,536 parts (as you've so kindly pointed out) and the 24-bit ruler is divided into 16,777,216 parts. But the way you describe it those parts would be evenly spaced, and they're not. Using out 100 and 150 cm rulers, the "top" 100 cm of the 150 cm ruler is divided into 65,536 parts, and the "bottom" 50 are divided into 16,711,680. All that extra resolution is used to describe the "bottom" 48 dB dynamic range.

                          The truth is in this example, the 16-bit AND the 24-bit scala represent the SAME audiosignal, just different scales.


                          Yes, but the problem is your example is totally wrong. The scale is the same. It's just that the 24-bit signal can "describe" a wider dynamic range.

                          A zero-value means silence, and the max-value means 0dBFS. Its the same wave, but once divided into 65536 parts (=16bit=2^16), and once divided into 16777216 parts (=24bit=2^24) .


                          No, that's not right. The max value stays the same...0 dBFS...but the zero value doesn't mean "silence", it means -96 dBFS in a 16-bit word or -144 dBFS in a 24-bit word. What does "silence" mean?

                          Some think the additional bits (24 vs. 16) would be added on top and therefore increase the headroom.
                          Only true when 16 bit data is converted into 24 bit data.


                          No, it's not even true then.

                          Some think the additional bits (24 vs. 16) would be added on bottom and therefore increase the resolution of silent pasages only (from -96 to -144 dBFS).
                          Plain Bull****************...


                          No, that's actually the way it is.

                          24 bit data has the same resolution as 16 bit and then additional 256 steps between EACH 16-bit value.


                          Hopefully by now everyone realizes this is wrong.

                          Isn't there also a benefit from higher resolution ??
                          Thank God it is!!!


                          Of course there is, there are several, but none of the ones you describe.

                          -Duardo

                          Comment


                          • #43
                            If you're like me who's stuck in between having a completely great system and some older semi pro stuff, it's worked out better if I record at 20 bit 48 k on my Adat, or 24 bit on my computer because I usually at some point I have to do some sort of analog processing depending on what someone has wanted. That way if it has to go in and out of AD DA processors a few times you'll at least minimize any sound degragation. I cringe when someone asks after I think it's mixed down okay if they could add some EQ or something even more whacked.

                            Comment


                            • #44
                              Originally posted by FanaticMusic
                              Is this "faux" 32bit?

                              Yes and No.
                              The data you record is rendered onto a resolution of 24 bits (still).
                              Which means, you just have 24 bits (3 Bytes) of data.
                              But the data is written "into a 32 bit-wide slot". (4 Bytes)
                              The reason for this is the way a computer threads the data. The adresses in memory, be it RAM or HDD, are used in Double-Byte Values (aka Words = 2 Bytes) or even as Double-Words (4 Bytes).

                              This means that the computer uses 32 bits (4 Bytes) of space for 24 bits (3 Bytes) of data. The only reason for it is the increased trasferring speed. You could say, that every 4th Byte is wasted cause its empty (8 zeros).


                              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?)
                              "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


                              • #45
                                I don't see this as a fight, my only intention is to inform the public about a common mistake.
                                So please don't take this to personal.


                                I don't take it personally. I didn't make this stuff up. And I too want to clear up some common misconceptions. So let's get to it...

                                Please read carefully now.


                                Believe me, I read all of this very carefully.

                                In the binary digitsystem as you surely know, one bit equals one digit.


                                So far, we're agreed.

                                So lets simplify the example and lets say we have the following:

                                .) A 2 digit datasystem ranging from 00 to 99. Let's asume this simulates the "16 bit system"
                                .) and then we add ANOTHER digit to increase the RANGE. ("the 24 bit system")

                                Like you also said, the digits will be added to the RIGHT of the numbers. And since we have an additional digit, but no data yet, the digits are zero "0".


                                So far, so good, except for that with digital audio we don't count up, we count down. That's why we talk about 0 dBFS and go down from there. "dB" doesn't mean itself unless it's referenced to something else...so when we're talking about digital recording levels we always state them as "-12 dBFS" or "-30 dBFS", where 0 dBFS is the loudest signal possible. It doesn't corrsepond to any particular volume level acoustically. When we talk about acoustic volume we always (should) express it in dB SPL, where 0 dB SPL is the threshold of hearing for the average human. Depending on how we set our levels, 0 dBFS may correspond to any SPL measurement. If we record a guitar amp with a microphone up close and set it up so it's right below clipping our converter, 0 dBFS may equal 130 dB SPL. If we're recording a small chamber group and we want their loudest peak to be righ below clipping on our converter, 0 dBFS may equal 75 dBFS. If we're recording someone speaking softly 0 dBFS may equal only 40 dB SPL. If we play all three recordings back through the same system without changing the volume controls, upon playback the peak of each signal will stay the same from one to the next. Are we agreed thus far?

                                I'll trim the next quote down a bit so we don't all have to scroll so far down:

                                So what happens In the 2 digit system it is 00, 01.....19, 20 etc...
                                when we now add the additional digit it becomes 00.0 ,01.0....19.0, 20.0 etc...
                                Now Look!!! We can now insert values between EACH value of the previous 2 digit system.
                                00.0, 00.1, 00.2....00.8, 00.9,01.0, 01.1....etc...


                                No, that's simply not the way it works. Where did you get that idea? If you want to illustrate it, again, we have to count backwards from zero, so...

                                In a two digit system it would go -01, -02...-99
                                In a three digit system it would go -001, -002, -003...-099, -100, -101...-99

                                So any number that can be expressed with two digits will be no more accurate when expressed with three.

                                Terrible analogy, as we'll find out below, but it comes a little bit closer than yours...

                                So your sentence "Hopefully by now everyone realizes this is wrong." seems not as if it would fit...


                                Again, hopefully now everyone realizes it actually does.

                                Your main problem is the confusion of dB values, which are indeed on a logarithmic scale, with SampleValues which are HIGHLY linear.


                                No, I'm not confusing the two at all. Sample values are indeed highly linear, and I never said they weren't.

                                For example a 24-bit Dataslot is indeed 1.5x the size of a 16-bit Dataslot.
                                BUT
                                The 24-bit DataVALUE-RANGE is 256x the size of a 16-bit DataVALUE-RANGE.
                                And you said the ruler has to be 1.5x times the length ????
                                The ruler of course represents the VALUERANGE and NOT the Dataslot.


                                Well, if you wanted your 16-bit ruler subdivided into the total number of values it would have to have 65,536 steps, and the 24-bit ruler would have 16,777,216 steps. Each "step" would be equal in length, so one ruler would be a lot longer than the other. You can see why it's a lot easier to think in terms of bits (16 vs 24) than values. Make sense?

                                For example in a 16-bit signed integer SampleValue World

                                0dBFS would equal "32768" and -6dBFS would equal "16384".


                                This is true, almost. To help understand why, I'll try to explain briefly. The first digit of a 16-bit word is called the Most Significant Bit (MSB). Thinking about a waveform as going up and down and up and down over a zero point, when the value is positive, the MSB is a "0", and when it's negative the MSB is a "1". So the "loudest" positive signal would be 0111 1111 1111 1111 (which is 32,767 expressed in base ten). A signal that's 6 dB down would be 0011 1111 1111 1111, or 16383. (32768, or 1000 0000 0000 0000, would actually be the lowest possible position below the zero crossing, which for all intents and purposes would be the same thing.)

                                Thus meaning between the loudest signal and its half are 16384 steps (32768/2).


                                Yeah, okay, worded a little oddly, but whatever.

                                In a 24-bit signed integer SampleValue World

                                0dBFS would equal "8388608"
                                and -6dBFS would equal "4194304". Thus meaning between the loudest signal and its half lie
                                4194304 steps (8388608/2).


                                Okay.

                                In the 24 bit system the same difference is rendered onto (4194304-16384=) 4177920 steps MORE than in the 16 bit system.

                                And I am sure you realized that the 6dB Range in the 24bit System contains 256x the amount of steps of the 6dB Range in the 16 System.(4194304/16384=256)


                                Okay, but what do those bits represent? Remeber, we're not going from 1111 1111 1111 1111 down to 0000 0000 0000 0000, but from 0111 1111 1111 1111 (as close to 0 dB FS as we can get on the positive side of the zero crossing) down to 0000 0000 0000 0001 (one "step" above the zero crossing) to 0000 0000 0000 0000 (the zero crossing itself) to 1111 1111 1111 1111 (one step below the zero crossing) to 1000 0000 0000 0000 (the maximum amplitude on the negative side of the zero crossing). As we move to the right, each bit repesents a value closer and closer to the zero crossing, which is why as we increase our bit depth we can capture a wider dynamic range...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. With 24 we can get closer and closer to the zero crossing and (theoretically) capture signals up to 144 dB down. Those extra bits are used when representing louder signals as well (as in your example), but since they're representing such a quiet component of the signal they make no audible difference whatsoever. Do you really think that when you're listening to something at a peak of 100 dB SPL that may get as quiet as 30 dB SPL the extra detail you've captured at -20 dB SPL will make any difference at all? You're claiming it does.

                                So please think again, and read my words once more.
                                Thank you.


                                Actually, I did...

                                And even when your VU- or Peak-meters show quite constant a value about the maximum, do you think the audiowave stands still ? It crosses the ZERO point quite often per second.
                                And when it comes close to zero everytime before crossing it, wouldnt that be a "silent pasage" ? Isn't there also a benefit from higher resolution ??


                                Again, are you saying that you can hear what's happening each time the audio passes through the zero crossing? That's where all your extra detail is, so unless you're listening to something that gets really really loud and then really really quiet, you don't need it.

                                -Duardo

                                Comment













                                Working...
                                X