Bad and Wrong Music Lessons: Getting Closer

By Shamus Posted Sunday Nov 11, 2018

Filed under: Music 27 comments

The last time I saw Close Encounters of the Third Kind was in the early 1980s. I was a kid, and I watched it on television. Some parts of it were too scary for me. The rest was too strange for me to understand.

It popped up on Netflix two nights ago and I fell in love with it. Spielberg was only 33 when he wrote and directed it. Most filmmakers have a rough early period of awkward films before they hit their stride, but apparently Spielberg was born a master of cinema. Must be nice.

The film has this iconic five-note tune that the characters use to communicate with the aliens. Even if you’ve never seen the movie, you’ve probably heard the notes. The movie was stuck in my head and as part of unpacking it I turned those notes into a song.

Every creative endeavor has its downside. Writing a story is fun, but proofreading is boring. Programming is fun, but debugging is a chore. Likewise, it’s fun to come up with melodies and rhythms, but mixing and producing sucks.

I’m sure you’ve noticed that it’s hard to proofread something you’ve just completed. When re-reading, you tend to read what you thought you wrote instead of what actually wound up on the screen. Likewise, I have a lot of trouble mixing something I just composed. I’ll hear all the various parts the way I composed them, and it’s not until I come back much later that I realize all the layers have blended together into mush. In particular, I love low frequencies and strong basslines, and often my kick drum ends up lost in the deep parts of the mix, like someone trying to clap out a rhythm in the middle of an applauding crowd.

I uploaded a spectrograph view of the track to YouTube so you can see what I mean. Check it out:


Link (YouTube)

This lets you visualize what’s going on in a sonic sense. Dots near the bottom of the view are low frequencies and they get higher as you move up the screen. The brighter the pixel, the louder the sound. If one area is saturated with near-white pixels, that means you’re probably losing some detail.

The kick drum appears at the 1:12 mark. You can see it’s a bright white blob of pixels at the bottom of the mix. In a lot of my older work, I never bothered looking at the spectrograph. If I had, I would have noticed that the bottom was already bright, so when the kick drum hit it couldn’t punch through the existing bass frequencies. You could kinda hear it, but in dance music it really needs to stand out. I mean, that’s sort of the entire point of the genre.

In this track you can see the bottom of the mix is basically empty aside from the kick drum. I added an EQ to strip out the low-band stuff from all of the musical tracks. This protected my drum, but I managed to mess up a couple of instruments and I didn’t notice until after I’d uploaded the track to SoundCloud and YouTube. At the start of the song it does this call / response thing where it plays the five-note motif high and then echos back a few octaves lower. This is something that happens during the movie and is a natural fit for the track. The thing is, the response is supposed to be this booming cosmic force. But I stripped out the low frequencies when I was trying to make room for my kick drum. I listened to the track, heard what I expected to hear, and hit upload.

There are a lot of other problems with this track, but this one bothers me because it would have been easy to fix. (What I should have done is disable the EQ until it’s time to add the kick drum. That would have left it with the intended booming response at the start of the track.)

Yes, I could got back to fix the track, but it’s always something. In figuring out how to avoid one problem, I blunder into another one. If I fixed this bug I’d just create another. It’s easy to pour a lot of hours into fussing with a track and I can never tell if I’m really improving the thing or if I’m just going around in circles. It takes practice to get good at the stuff and I’m just not there yet.

Despite all the bellyaching and navel-gazing, I actually dig this one. Here’s the track view if you’re curious:

Getting Closer
Getting Closer

Yes, I’m aware that Deadmau5 did the same idea years agoIf that track is region-locked for you, search for “Closer” by Deadmau5., except his take on the melody is awesome and actually works as an EDM track. Still, projects like this are a good learning experience for me and maybe someone out there will dig the tune.

 

Footnotes:

[1] If that track is region-locked for you, search for “Closer” by Deadmau5.



From The Archives:
 

27 thoughts on “Bad and Wrong Music Lessons: Getting Closer

  1. CJK says:

    You’ll find that spectrograph a lot more useful if you switch it to a logarithmic scale – since each octave is a doubling in frequency, viewing a spectrogram linearly distorts the relationship between notes and wastes a lot of space. The entire top half of that display is pretty useless since music has little content above 10kHz and virtually nothing above 16kHz, and the bit you talked about in this post is all in the bottom 50 pixels.

    I’m not familiar with that software, but it’s almost certainly an option in the context menu of the y-axis.

    1. Echo Tango says:

      The spectro-whatever at the bottom of the screen, could use a better X-axis scale too. I can see the low-frequency stuff all takes up many inches on the left of the screen, but the high-frequency stuff is all squished onto the right of the screen.

      1. CJK says:

        Yeah, but that’s actually correct – it’s much more useful this way.

        You want each octave to take up a constant amount of space, so the space from 25Hz – 50Hz should be the same as 50-100, 100-200, 200-400 and so on. It gets “all squished up” on the right hand side but that’s fine – what you’re interested in is notes and octaves, and they’re represented consistently all the way across. More resolution in the higher frequencies isn’t useful – there are still only so many “correct” tones in the scale, you don’t get extras to fill in the space higher up.

  2. It’s the old saying, less is more. I.e. if you add something, ponder if you should also remove something.

    Also note (pun intended) that sometimes the best way to make something louder is to “reduce” the volume of everything else other than the thing you want louder. This preserves dynamics and you avoid using compression to squash things down.
    Sure the listeners may need to turn the volume knob up on the track (since it would be defying the loudness war).

    1. Echo Tango says:

      Does most commercially-produced music include normalization info nowadays? (Max volume, min volume, volume info for a track, volume info for an album) I was reading about this back in college (so like, more than a decade ago now) and it seems insane to me, that all vehicle stereos, phones, and portable music devices, don’t have built-in options to let you specify the loudness range you want, to listen to your music. Something like, keep it above volume X so nothing’s imperceptibly quiet, but keep it below Y so I don’t damage my hearing.

      1. Myself I target -23 dBFS RMS for my music and audio work. The internet radio station I’m the techie at targets -23 LUFS (almost the same as dBFS RMS).

        Youtube I think targets -16 LUFS or -14 (not sure). iTunes I think is -16 LUFS.

        As to music including normalization info. I don’t think they do. I only vaguely seem to recall movies with Dolby sound having a loudness for dialog that you can tweak in your home theatre (other standards like dts may have something similar).

        It’s possible some HiFis or car stereos has a “Night Mode” that compressed the highs so you get less dynamics.

        The issue though is that one you have compressed the loudness you can’t get it back. If you have a very dynamic track you can always compress it “live”, or adjust just before playback by using loudness meta info if present.

        Software players like Foobar2000 support Replaygain (later versions of Foobar uses european broadcast standard EBU R128 for loudness scanning) and sets the playback volume per track at playback but does not Manipulate the audio at all.

        Sadly no CD music (and few digital online retail music) has loudness info or follow any standards. Exception being services like iTunes or Spotify (which should).

        Replaygain supports peak scanning and LUFS calculation (track and album), if you play tracks from multiple albums then track mode is best, if you play albums back to back with seamless track transitions then album should be better.

        Media playback software like Potplayer or VLC and so on has normalization, but I often find that to cause a volume pumping effect (there’s no loudness metainfo so it’s just compressed “live”) which is kinda silly.
        I don’t think purchased downloadable tracks from iTunes or Spotify has Replaygain info. iTunes tracks should have some Apple specific metainfo though.

        The main issue though is that songs delivered to commercial radio still do the “loudness war” nonsense. Which is ironic as at least one large radio station in Norway uses -16 LUFS (per EBU R128 standard) as their level and plan to slide that down over time down towards the EBU R128 target of -23 LUFS.
        Which means tracks that are hyper compressed will actually be played back more quiet than a highly dynamic track which is rather ironic.

        There is a limit to how dynamic you can have a song though. Humans seems to prefer a average loudness range of max 30 dB. So a range of 0dB to -60 dB should be plenty. Do note that this is average loudness not min max peaks.
        Also note that a movie theatre (cinema) uses 83 dB SPL as it’s max loudness. So a CD with it’s 96 dBFS (or 95dBFS if we throw away the least significant bit as that is usually just dithering noise) has enough volume range.

        Movie theatre loudness of 83 dB SPL refers to a RMS average (I forget if it’s dB(A) or DB(C) weighted, myself I prefer “Z weighted” which is a flat filter).
        So with a 30 dB range, that would mean 68 dB SPL to 98 dB SPL. 83 dB SPL is deemed safe for 8 hours I think (can’t recall), and anything below 77 dB SPL you can probably listen to 24/7 without risking permanent damage to your ears.

        I’ve found that I like to naturally listen to audio on my PC at the volume of -15 dBFS to -30 dBFS (averaging around -23 dBFS), and that this translates to 60-70 dBSPL (sound pressure) to me. Which is good since I sit with my headphones on 99% of the day.

        Do not underestimate the loudness of headphones, some have a sensitivity or SPL of up to 98 dB SPL, and listening at max volume with that could cause permanent hearing damage after just a few hours if unlucky. Do note that this is average, peaks can be higher but not too high. Peaks at above 120 or so dB SPL can cause instant permanent damage.

        My info is pretty anecdotal now, I’m sure there are beginner guides with up-to-date info on this out there that are better than my ramblings here.

        1. Echo Tango says:

          I meant on-the-fly compression, not baked-in stuff. That’s actually part of the problem this is trying to solve – if one song on YouTube is too loud, one song is too quiet, or one song has no dynamic range because it’s been compressed into a small range of loudness, those are all problems. :)

          1. melted says:

            Right? It seems like you should at least–especially on a computer!–be able to tell it, “by the way, never let anything be louder than x dB, even if other volume controls indicate otherwise” so you don’t get the thing of turning the volume up to listen to something quiet, and then forgetting you did that and playing something normal-volume or louder after it.

            1. Aitch says:

              That’s why I love having this sound card. One of the options in the Control Panel is “SVM”, Smart Volume Management. It does realtime compression / expansion for audio output, which is an absolute godsend for everything that isn’t music. Quiet videos get louder, sudden loud things are kept level, and it’s fast enough to not cause any kind of annoying volume bouncing you might get with something like a straight compressor.

              Listening to music with it on is only good if you’re looking to keep things quiet on a low volume setting. Otherwise it does what you’d expect and totally flattens the dynamic range – and most modern music already pushes that to the limit as is. See Also : The Loudness War.

              The card is a Creative X-Fi Titanium HD, but I’m guessing similar features exist on other cards. It’s worth every bit for me just on a personal enjoyment level having non-software digital music decoding (or at least processing) with a suite of tools like a graphical eq, 3d audio modeling, “crystalizer”, etc. I never thought much of it, but this stuff really saves me a tremendous amount of annoyance with how variable something like YouTube videos are even compared to TV (and we’ve all been woken up by that 4am commercial that’s somehow 3 times louder than anything else)…

  3. CJK says:

    One technique you might want to look into is “sidechain compression” – you can use the loudness of the bass drum track to control the strength of a compressor / limiter on the bassline track(s) (or indeed any other effect, like your EQ), so the kick drum will “punch a hole in the mix” on each hit.

    1. CJK says:

      I should note that this is a technique to apply gently – pushing it hard is what gives a track that slightly queasy “pumping” feel that plagues EDM circa 2000. But in moderation it’s a great way to get the bassline out of the way for more kick impact.

      1. Echo Tango says:

        If you want your music to induce vomiting, as part of the artistic experience…?

        1. Arkady English says:

          I believe that’s called [insert-name-of-first-artist-to-hit-mainstream-popularity-after-you-turned-30-here].

  4. BTW! Shamus. That deadmaus video is unavailable. Comments are off, there are some likes and dislikes and 6526 views. No idea if it’s temporary or if it’s deleted. Doesn’t seem to be a DMCA or regional block either.

    While I’m off topic. Seen these?
    Obsidian anouncement on joining the Microsoft Studios.
    InXile anouncement on joining the Microsoft Studios.

    I wonder if Microsoft will hire Chris Avellone and he’ll end up bouncing between inXile and Obsidian (studios are like a 15 min drive from each other).

    Would also be fun to see inXile and Obsidian colab on a large project in the future (kinda like how Rockstar’s studios colab on Red Dead and GTA).

    I’m looking forward to what the devs can do with a solid financial backing and QA and resources that allow higher production values.

    Obsidian and inXile’s titles are the perfect catalog filler for Microsoft, though I do hope we’ll see PC releases of games on GOG as well.

    1. Echo Tango says:

      I can listen to the video just fine; Region-lock? (I’m in Saskatchewan)

      1. The “Deadmau5 did the same idea years ago” link? It still says video unavailable to me.

        1. Syal says:

          The deadmau5 video is playing for me in the US, and shows 4 comments.

        2. Echo Tango says:

          Yeah, the fifth-deceased-rodent video.

  5. Joshua says:

    I had thought this Close Encounters idea had been done back in the 90s with a set of albums called Happy 2B Hardcore. Looked it up, and it turns out it was actually the music from 2001: A Space Odyssey.

  6. anaphysik says:

    Spielberg was only 33 when he wrote and directed it. Most filmmakers have a rough early period of awkward films before they hit their stride, but apparently Spielberg was born a master of cinema.

    In fairness, by that point Spielberg had been making home movies for about 20 years, amateur features for about 13 years (starting with a Close-Encounters-esque movie), and “real movies” for about 6 years (or if you don’t count TV-movies, about 3 years, with CE being his third theatrical, and Jaws his second). Mad talented, yeah, but continuous dedication is a big deal.

  7. AzaghalsMask says:

    @Shamus: Looks like the track on Soundcloud is not enabled for download. For those of us who like it, could you change that?

    1. Shamus says:

      Oops. Fixed.

      SoundCloud always defaults that to off and I never remember to change it.

  8. marty says:

    The problem with using the spectrograph is that it’s displaying information from the DAW and not what is coming out of the speakers, not what’s hitting your ears. Same thing with any meters on the mixing board. And everything in software is assuming a flat frequency response (all frequencies being reproduced with equal accuracy) that, unless you’re dropping some decent money on monitors, amplifiers, and acoustic treatment for your production studio (see: bedroom, garage, basement, child’s bedroom that you were allowed to put a desk in after they went to college), you probably don’t have. Also, ears don’t have a flat frequency response curve either. The spectrograph isn’t a ‘lie’ but it’s not really accurate.

    You can safely assume that your signal chain is going to significantly attenuate everything under 10Hz, so putting a highpass filter on the kick so that the DAW doesn’t factor this into mixing is a decent enough idea if you want the kick to stand out more by making the channel itself louder. I think 50Hz is generally considered where the kick ideally sits but you can mess with the filter so that it sounds right while giving yourself some headroom in the DAW. A little bit of fuzz/distortion from a plugin can go a long way in adding harmonics to your kick if it’s bass-y enough but doesn’t cut through the mix. Using multiple kick samples is another good way to make the kick pop. Consider that engineers recording real drums will mic the batter and resonant heads of a kick drum so they get enough of the attack and ‘whomp’ for mixing.

    tldr, mix with your ears. you did your filtering backwards (also, eq the bus either not at all or with the lightest touch–if you’re doing a lot of eq on the bus, you can kinda assume the problem won’t be fixed by doing that).

    PS. I’m still waiting for a series where you try to make procedurally generated music.

    1. Echo Tango says:

      Even the musicians aren’t safe from the robots – bring on the computer-generated music! :P

  9. I found the dedmau5 Closer song here https://www.youtube.com/watch?v=eI0KIsRo1zk
    Comparing it to Shamus’ song the bass is dramatically different.

    Closer has a sliding bass (bauuuum). Getting Closer as a tap bass (bump) in comparison. It’s easy to think a bass is only lows, it’s usually not. Also Deadmau5′ bass is a layerd sound with 2 or 3 bass sounds and later possibly a 3rd or 4th layer.
    Whomever out up that video used a frequency bar thing so you can see the bass slide from quite high to low then the “kick”. There may also be some ducking of the other sounds during the deeper part of the bass and the kick to make it seem even stronger. And he may have placed the notes so they do not overlap on the kick itself too much.

    Deadmau5 also has instrument placements (as a orchestra would) and not just cross/stereo reverb. There is also a lot more audio “behind me” so some of his sounds are out of phase/inverted placing them in the surround. So his sound stage is really wide and wraps around you.

    One interesting thing though. Deadmau5 did a different version earlier https://www.youtube.com/watch?v=tmSLZuVJP2o
    Which sounds closer (pun intended) to Shamus’ “Getting Closer”. Deadmau5 “beefed up” his final cut/release?

    @Shamus this seems to be the official “Closer” YT vid link https://www.youtube.com/watch?v=Pd2QAy59IlY
    Not as nice graphics though.

  10. Linus Tech Tips did a short video on Deadmau5 music setup (they where invited) https://www.youtube.com/watch?v=dBiqFNNfudA

    Hmm. Am I the only one that thinks the game design of Good Robot redone to use music by Deadmau5 and starring erm “Deadmaus5” could be awesome musical game?

    When I saw the lights and the music in this video https://www.youtube.com/watch?v=plywqXrNtJI
    I imagined how cool it could be to play “Deadmau5” speeding through levels like a supermouse, dealing with music inspired enemies while Deadmau5 music is playing. A few special levels could even be synced to the music (would work best if these where autosrolling and enemies emerged/attacked in sync with the music).

  11. germdove says:

    Soundcloud is so incredibly frustrating with their embedded player. All I want is some volume controls. They already exist in the widget, there just isn’t any way to use them by default. So now I’ve had to spend time coding up a userscript to give myself volume controls because they decided users wouldn’t make use of it.

Thanks for joining the discussion. Be nice, don't post angry, and enjoy yourself. This is supposed to be fun. Your email address will not be published. Required fields are marked*

You can enclose spoilers in <strike> tags like so:
<strike>Darth Vader is Luke's father!</strike>

You can make things italics like this:
Can you imagine having Darth Vader as your <i>father</i>?

You can make things bold like this:
I'm <b>very</b> glad Darth Vader isn't my father.

You can make links like this:
I'm reading about <a href="http://en.wikipedia.org/wiki/Darth_Vader">Darth Vader</a> on Wikipedia!

You can quote someone like this:
Darth Vader said <blockquote>Luke, I am your father.</blockquote>

Leave a Reply

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