1. 43

  2. 3

    As always, great explanation of various technical details. Great talk as Rust introduction.

    1. 3

      Nice talk, but the mic constantly auto-leveling was very distracting. oof.

      1. 5

        Ugh, yeah, I’m aware. I tried running it through a normalizer, but to no avail.

        1. 7

          It sounds a bit weird but running it through an expander (to get quiet parts as loud as the louder parts) followed by a really aggressive compressor (really high ratio, to smooth out volume spikes) might do the trick.

          1. 9

            I know this is getting off topic, and maybe this is a weird question, but: how did you learn the knowledge required to make a suggestion like this? I’ve tried somewhat half heartedly learning more about audio, but always wind up baffled by the complexity of the interface for software tools in this domain.

            Tangentially, which tools would you personally use to implement your suggestion?

            1. 19

              There are plenty of people who are way more knowledgeable than me at this, but I’m happy to share what I know. I’m actually going to answer your questions out of order because one of the somewhat less interesting answers ended up being really long.

              Lots of this ended up being a ramble to get my thoughts out, but if you’ve got other questions, you’re welcome to ask them.

              Also, reasoning behind the expander/compressor (actually might be better to just severely boost the input signal and add a compressor, maybe with a limiter as well, but the effect will be similar) rather than a normalizer is because normalizers generally operate on the whole track at once (It’ll bring the highest peak up to a specific volume) while a compressor will operate on a slice of audio at once. Anyway the issue is with the dynamics, so the goal is to tweak them to make it better.

              EDIT: after playing with it, just boost the input signal and add a really heavy compressor with a limiter. The expander was actually not a good idea.

              always wind up baffled by the complexity of the interface for software tools in this domain.

              This was a huge problem for me starting off. It’s very hard to explain over text, but I think this is something you can only really learn by experimenting (though if you’re specifically looking for something to learn with, Mixing Audio: Concepts, Practices, and Tools and Mixing With Your Mind are both very good books for learning mixing theory). Most software tools end up being pretty close to their analog counterparts. DAWs usually have a set of faders and an EQ, with a place to add additional plugins.

              Same as software development, start with using only the portions you know and start adding more and more complexity as you learn it. Levels/gain and EQ should probably be first, followed by compression. Once you learn that, there’s a ton of other stuff out there.

              which tools would you personally use to implement your suggestion?

              In general I like using Logic Pro X. The built-in plugins are pretty good quality wise and it’s reasonably easy to use. Reaper would be another recommendation, especially if you need something cheaper and/or on multiple platforms.

              This one’s a little different because it deals with video as well. I’m not a fan of Final Cut Pro overall, but I’m pretty sure there’s a way to use that for the video and deal with the audio in Logic Pro X.

              I know open source tools exist (I think Ardour is the main one), but I haven’t found any that can measure up to the same quality of Logic or Pro Tools (or even Reaper). If you’re looking for something smaller to play around with, Reaper is fantastic, cheaper, and has really good platform support. I’ve tried moving back to Linux for my laptop (I ran the Linux Users Group in my college and it’s still something I’m passionate about) but because the quality of audio tools (or at least the audio tools I know - there’s also pro tools on Windows, but I’m not a fan of that either) is so much better on MacOS, I’m kind of stuck here.

              how did you learn the knowledge required to make a suggestion like this?

              1. Experimenting and playing around. Back in high school, I used to make super low quality recordings of a band I was in at the time - this consisted of hanging a microphone from a hook in my parents’ basement and manually setting levels by moving it closer or farther away from some instruments. This eventually moved on to getting a small mixer and managing separate microphones but recording the mixed down version rather than separate components.
              2. I never ended up finishing it (because I just couldn’t finish Music Theory II) but I have about 2/3 of a “Music Technology” minor (which is essentially Sound Engineering lite - I say lite specifically because there’s a lot of stuff I don’t know that a full Sound Engineer or Mixing Engineer would) along with my Software Engineering degree. Specifically, there was a class where every week or two they gave us a bunch of raw tracks and told us what to watch out for (this class was incredibly helpful - to be honest a lot of it was learning hacks to make things work). We had to come back with a rough mix.
              3. Experimenting and playing around. After college, I helped run sound for a few churches. It’s not always the most interesting (and I know plenty people aren’t a fan of churches in general - if you have ethical issues with this, there are other options, but they’re harder to find), but you learn quite a bit when you just have to make something work.
              4. For the last 2 to 3 years or so I’ve also been helping run live sound at a swing dance every week. This started off as just loaning some equipment I had lying around but turned into me filling in whenever other people couldn’t make it and eventually being one of the main 2 people who run sound there. I tend to be more interested in recordings where I can tweak things after the fact, so I try to focus on live multi-track recordings and the other people tend to focus on live sound.
              1. 8

                Really interesting! Thanks for the lovely response. I’ve added those two books to my list, so if I ever decide to dive back into this, I’ll have a better place to start. :-)

          2. 1

            Fyi: I’ve downloaded an mp4 of the video, threw it inside kdenlive, split audio and added “AutoCorrection->Normalize”. Now it’s equally loud, except a little too loud :P

            1. 2

              That’s interesting — when I tried normalizing, it didn’t help at all, probably due to what was discussed in the other comment thread about normalizing usually being applied uniformly across the entire audio. Maybe kdenlive does something smarter?

              Also FYI, I now have a lavalier mic that I’ll use for future recordings \o/