Digital Audio For Geeks

http://www.youtube.com/watch?v=d7kJdFGH-WI

This video, via Xiph.org, offers a deep exploration of the multiple facets of digital audio signals and how they really behave in the real world. As a bonus, it’s done with old-school analog gear!

Host Monty Montgomery continues the “firehose” tradition of maximum information density from his earlier video, A Digital Media Primer For Geeks. He covers sampling, quantization, bit-depth, and dither, using both modern digital analysis and vintage analog bench equipment.

Check it out and let us know what you think!

This work is free-to-share under a Creative Commons license.

30 thoughts on “Digital Audio For Geeks

    1. Why so divisive?

      Information like this should be neutral ground.

      Really, what made you react to such a cool video with an insult to people who play analog synths? Did you just see DIGITAL and jump at the opportunity?

      Just curious

      1. to be honest with you its the derisiveness on this site i am curious about. i honestly thought that my post would have been hidden by now, interesting that as of my writing this it’s 10 for 11 against, much closer than i would have expected. i’m a programer and sound designer, everything is always an experiment and yeah i’m a bit of an a hole.

        by the way i use analog stuff as well as digital stuff for 34 years now, after all they’re just tools to make sound with. neither right nor wrong, better or worse…

        1. on this site? you should have said something about how much macs suck dick, and how ipads are not the future, they are just toys, etc.

          protip: that will do it

  1. Excellent video!

    No need for analog vs digital divisiveness, though. We all know digital and analog can both can be great for audio and for sound work – when used for the right things. I’ll take a digital reverb over an analog one, and a convolution reverb is better yet!

    It’s hard to beat real analog sound. But, as Dave Smith is demonstrating, a best-of-both worlds approach is hard to beat!

  2. I completely had forgotten about the Emagic audio interface shown in the video. It goes back to 2000, or so, I think. I have to admit I got completely lost at times in the video, but, I took away that we probably don’t need to be doing some of the things we are doing with digital recordings. I did find it fascinating to hear that a cassette at best could be around 6 bits, and a reel-to-reel around 13 bits. p.s. I loved the vintage shirt!

    1. The bit depth comparison between Digital and analog is the most false and misleading statement I can imagine. And nowadays it’s even spread via Wikipedia…please stop this !!!!
      No…a cassette definition can not be compared to 6 bits digital !!!!
      6 bits means 64 step definition while a cassette has INFINITE steps definition.
      Yes analog is INFINITE.
      The only comparison can be made to the signal to noise ratio.

      1. While you’re right that tape recordings are not stepped in the same way, I think the point was more in comparing the upper and lower limits of resolution, not in between. To use a metaphor, I think you’re confusing a discussion about which ruler is longer for one about which measures in the smallest increments.

        1. quote :
          “..tape recordings are not stepped in the same way….”

          They are NOT stepped !! fullstop.

          64 steps (6bits)….huhu let me laugh…..

          1. First, you’re just splitting hairs for some reason, I think its apparent what I meant.

            Second, your insistence on this infinite resolution thing is really fucking annoying. If you really want to get as specific as possible, everything is stepped at the level of the electron, infinite resolution is simply impossible due to the nature of physical matter.

            1. Actually, if we are describing waves, as audio is, there is no stepping because its almost impossible for a membrane (auditory or source) to fluctuate in a stepped fashion.

      2. Cassette tape is limited in many ways.

        The physical characteristics of the tape limit the range between the noise floor and the maximum volume. The width of the tape limits the accuracy of reproduction. The speed of the tape limits the taped ‘resolution’.

        Suggesting that analog doesn’t have significant limits is ignoring reality.

      3. You missed the point completely. The signal that comes out of the D/A converter is the same for a normal continuos cassette and a 6 bit digital one. You have no way to dscern them as long as you don’t go beyond 20kHz signals. The continus resolution of analog gear does only add quality to a signal beyond the frequencies a human ear can hear and are filtered out anyway long before they reach the cassette or amplifyer.

        1. sorry Saxifra you can’t even make the difference between bit depth and sampling frequency….better go to school and study electronic (as I did)…..btw analog cassette will never go beyond human hearing frequency because cassettes are limited to 12/15 Khz max frequency…that’s a fact…as opposite to the 6 bit limitation wich is all the way false……
          Try to pass your audio thru a bit crusher and you will understand…..64 steps I told you….

          1. Oh, I get it now, the “analog 4 lyfe” crowd are old washed up EEs who feel threatened by the digital revolution (which seems ancient history to everyone except them).

            Sorry but driving one of those cold and dirty “cars” that are propelled by a series of small explosions will just never be as satisfying as riding in a carriage behind a living, breathing and crapping horse! It’s a much more “warm” experience! As president of the Buggy Whip Manufacturers Association I assure you my opinion is unbiased!

      4. If you’re still talking about stair steps, and claiming that a casette has infinite resolution, then you either haven’t watched, or haven’t understood this video.

      5. Re the 6 bit cassette he was talking about signal to noise ratio NOT any other quality of the sound AFAIKS. There was no attempt to mislead there.

  3. Great video! Hopefully this will dispel a lot of digital audio myths and help vaccinate those in the market for digital audio interfaces against some of the more deplorable marketing practices going on today.
    The real difference between the audio quality of a great audio interface and a stellar mastering-quality one really lies in the clocking, which mostly results in phase disagreements (but these are, again, pretty small!). I’d love to see a video that goes into digital phase effects!

  4. I feel like a total nube. Then again he said he picked his examples carefully. We’ve all head what 8bit sampling does to music. IMO compact cassettes sound far superior to 8bit recordings though according to him it should be slightly the opposite. I wish he could explain why that is. Chime in If you know.

  5. I guess that there are people that think that the differences have something to do with stair-stepping and noise, but that doesn’t mean that the differences in practice are never real. 🙂

    The differences between analog and digital gear seems to have everything to do with the kinds of algorithms that get used in computer code, and the kind of interfaces that get put on the end result. If you have code that manipulates a digital signal in the time domain, it is often an order of magnitude cheaper (in cpu cycles – it may be the only feasible choice given the CPU budget) to do things like taking every n-th sample where n is a floating point value and do interpolation between samples or just take the closest value. A higher sample rate will also give more FFT bins to work with (ie: give me a 10 octave range with 1 cent resolution per FFT bin and I have an algorithm that’s easy, if I have to set multiple neighboring bins to get a reasonable approximation, then maybe that same algorithm becomes infeasible).

    The analog to digital conversion ensures that there are no frequencies beyond the limit on the way in to digital, the digital to analog will convert the samples to a smooth wave on the way out… but if in the middle of your signal processing, aliasing gets introduced, it can never be removed. Algorithms on the input don’t run in constant time (ie: O(1) per sample), so there will be some cases where the analog implementation may well be superior.

    And in computer science, there are people that believe that anything that you can write in Python you can write in Assembly language… It’s not really true because everything that gets made in the real world is influenced by what is hard and what is easy; and some products will just never ship if they can’t get something working within the resource constraints of people and materials.

Leave a Reply to Rob Cancel reply

Your email address will not be published. Required fields are marked *