TR909 Have I Discovered its Secret Ingredient?

[@morphoice]

The slight delays cause phase differences which make it sound organic and not super repetitive. That's why, as a plugin developer, I include those subtle nuances in my drum machine plugin. Currently modeling a LinnDrum from my hardware and there are sloppy delays of up to 7ms

[@Rompler_Rocco]

I want to go around today playing the two examples to strangers. If they can describe the difference, I'll give them $50. But if someone punches me in the face first, I'll allow myself to purchase a Hydrasynth.

[@StarskyCarr]

Winner - you’ll have the Hydrasynth within 10 minutes I reckon 😆😆🥊

[@notsure1135]

Vital is free

[@MwahMwahProductions]

I hope someone gives you a good wallop! 😂

[@DerDanko]

The discrepancies in timing might help other sounds to come through with their transients. When I finally had sample accurate timing in my DAW, I noticed a lot of transient masking and it took me years to figure out sometimes a track or instrument popped a lot more when moved a smidge later on the timeline to be de-masked by the major sounds like kick, snare and so on. I still do that to this day. Cranking an enveloper or compressor for transients and boosting high mids hit the dynamic roof and still did not allow certain sounds to have their transients audible because of the accurate timing, and so the solution was to not use compression or EQ and instead just slide things a bunch of ticks.

[@notsure1135]

I bought battery 3 about 15 years ago and the manual said ‘push things around by a few ticks to have the transients pop out.

[@StarskyCarr]

Yes its a little trick I've used a lot... slightly delaying a snare so the transient of the kick pops out and isnt affected by any phase relationships with anything else.

[@DerDanko]

@@StarskyCarr Rock solid timing is overrated. :) As for the TR-909 Shuffle/Swing "algorithm", people at the time were saying it had some magic just like the MPC 60. I was in the studio around 1995 together with Antilloop among others at the time, and they had gotten to borrow the TR-909 from Ola Håkansson (the band Secret Service, founded Stockholm Records and much more). We captured the different Swing settings into Cubase and compensating for margin or error (this was, after all, MIDI) it looked just like simple percentage steps on the off beat and that was that. Just like your analysis indicate. So we never bothered to use those Groove Templates we had just spent time on.

[@MrOtistetrax]

@@DerDanko Spot on. Pretty sure you can actually hear that effect on the quantised examples he plays.

[@phurniture]

I do this on my Digitakt (an even in the DAW) when I have two sounds hitting on the same note. I typically keep things like kick drums unaffected, but if there's a percussion sound that lands on top of a kick I might move it a bit later. I almost always move claps that land on the same note as a kick a bit ahead or a bit late which gives the perfect feel IMHO.

[@coolo73]

Out of curiosity, I wonder what the machines were clocked from and how that may have affected the relative timing. Back-in-the-days when I first started making music, we would daisy chain drum machines, particularly using a TR-909 , 707, or 727 as the master clock. those units had both MIDI and Roland Sync-24, so we could run the TR-808 and whatever Yamaha or Korg machine all together. Imagine 2 or 3 guys each sitting with a drum machine in their lap, improvising together like a electronic version of a drum circle. Experiencing that at the age of 14 is probably why I'm still obsessed with drum machines to this day.

[@JayKaufman]

A worthwhile asterisk to this. CV is notoriously much tighter sync-wise than MIDI - as is audio-based sync. If you're slaving something from one of those clock sources, you're more likely to get more solid MIDI timing than doing so from say, a DAW - which notoriously has terrible MIDI sync abilities (although certain DAWs are better than others).

[@rv0soft]

This. As far as I could tell, there is no info given about how things were synced to the DAW. Which leaves me with many questions.

[@wylatron]

Yes, that would change perception. Being a bit late on the beat makes things feel laid back. Being a bit early on a beat makes it push and propel the track forward. When you hear people talk about bands being ‘in the pocket’ and ‘groove and feel’ this is what they are talking about. What a marvellous find, and a window into what makes this machine one of the very most loved world wide❤

[@3dsman]

Well said. I agree. I wonder if they intentionally 'skewed' any roundoff errors to always err in the direction of being on the early side as that would generally be more desirable?

[@thefishdog]

the original 909 shares stuff in the circuits and stuff phases , its the limitations the engineers had to create a drum machine in a price category back then, they did a good job

[@TobyBorrow]

The 707 does a quick catch up flurry every 6 bars. I never noticed it till it was pointed out. It is the nature of the beast to be a little sloppy

[@AlexBallMusic]

This is the good stuff. The E-mu SP-1200 is like this. I did an equally nerdy experiment where I played it in from its own clock. I then tempo matched my DAW to the recording and sent the clock back to the SP-1200 and the hits all landed differently on the grid with all sorts of variation and seeming randomness. This is even with a unit that isolates the clock so there's no jitter at play. I'd heard about it, but it was interesting to confirm it. Then there's the MicroComposer clocks, which is a whole thing too. Nerds unite. 🕑🤓

[@StarskyCarr]

haha... I know its crackers that we end up looking at this stuff... but its the 'why?' we keep on coming back to. .. and obviously 99% of the population would describe us as deranged.

[@ghfjfghjasdfasdf]

Great stuff, from both of you!

[@kaitlyn__L]

This has validated the way I almost never put anything except kick 100% on the beat when I'm manually sequencing beats. I don't think too hard about it, I just wiggle them around with the left and right arrow keys (while 100% zoomed in) until they sound "right".

[@itsraininfire12]

As someone who’s been playing drums professionally for a long time, I was astounded at the difference between the two examples. You aren’t hallucinating at all

[@brianlespoir6287]

Use those audiofiles to create your own quantize settings, I did it with my 808.

[@fullytrapped.]

3:18 you're not hallucinating but you're glowing like the morning sun, got my welding goggles on for this one 😎

[@StarskyCarr]

haha. I think I'm not actually touching the chair either... floating a fraction of a millimetre above it, but imperceptibly floating nonetheless.

[@PBHorrorGaming]

Interesting, that is a major difference .

[@phtamas]

I don't own a 909 but experimented a lot with adding a very small amount (must not be audibly off grid) of timing randomization to hard quantized electronic drums and it seems to have some psychological effect. Slightly randomized patterns won't sound more "groovy" or "human" at all (in fact they're indistinguishable from their hard quantized counterparts when played only for one measure) but they can be repeated for quite a long time before they start to feel boring, unlike hard quantized ones which need frequent variatons/fills to maintain interest.

[@JonMurray]

So many awesome tracks created with that special sauce we never even knew about!

[@daemmerung]

Isn't that the same phenomenon as with the TB-303? I think it's the secret ingredient that makes the sound of the original so unique.

[@catiec4ke]

Great video Mr S. I noticed the same thing when I did a comparison between the the TR-09 and an MC-500 triggering the 09. The MC had an undeniable feel to it that was 'all over the place' in Logic. So you're not hallucinating. As far as I understand it, the processor in the MC-500 is from the same lineage as the 909 and the Roland engineers were squeezing everything they could from it. I think it's a really interesting topic.

[@van_daddy_z]

Interesting. When you look at the production of classic electronic tracks made in the 90s its seems the majority used a one-shot samples of 909s rather than the 909 itself and wonder if the timing thing is a red herring.

[@StarskyCarr]

I wonder. Loads where made with samples. I made a fair few myself 😆but I’m blaming them not becoming classic on the timing of the drums! Having said that perhaps the classics created a genre because of the vibe… I do remember wondering why my beats never quite hit the mark I was looking for. Interestingly the only one I was really happy with used another track as the template for the drum rhythm. I converted the audio to midi. Only remembering this as I write. Doubtful but a fun techy ‘conspiracy’.

[@paradoxstate5018]

As someone in the comments have said it was partly due to what they had to work with when the '909' was created, you'll find it all over 'vintage' electronics not just the '909' its the sole reason it all sounded like it does, while I am eagerly awaiting Behringers RD78, any 1 on 1 comparison will show that something is missing.

[@Robguitarjarvis]

This does make a difference, I always try to emphasise these little quirks to my students,( I’m a guitar teacher ) and being just behind or in front of the beat definitely gives a different feel to the whole work whatever the style.

[@kemek3000]

Back in the day I used a 707 as the master clock for my studio (so I could sync a 303 as well). The timing was all over the place.

[@tobypitman]

I remember making a load of groove templates for Logic using the MPC300 swing quantise settings back in the late 90's with Recycle. I think I used a 1 sample click to do it so it was pretty accurate. I remember I was quite surprised to see how off grid the timings were for all the percentages like you see here. Stylus RMX has a bunch of Groove Lock quantise setting based on the 909 and MPC all of which have these subtle timing fluctuations.

[@StarskyCarr]

Yeah I did this using a track I loved years ago and used the template an awful lot. It only had a single closed hat skipping just off beat z but really created a beautiful groove. And… Stylus RMX!!! I used to use that on everything , and completely forgot I had it. I had loads of beats I loved that software. I’ll have to install and have another play.

[@bob-rogers]

Cool! I love stuff like this. To me, the non-aligned clip sounds better. Kind of more open.

[@djtorule]

Isn't this awesome, it's like the universe, those slight imperfections is where the perfection is. Maybe the 8 bar loop was short, but when it's quantized it sounds (feels) bland, but the slight variations subconsciously adds character. Great analysis & observation.

[@TrymoFilms]

A long time ago I read an interview with a musician talking about this - his view was that the 909 had some sounds (maybe the snare) that hit early and the 808 had some that hit late. But you’re telling us it depends on the swing. More analysis is obviously needed. Do all 909s vary by the same amount on each sound and % swing?

[@slavikdoter]

this sound comparison made me smiling at the first sounds of 909

[@moorestreetfootscraytrucks2909]

Very interesting 🤔. Brilliant investigation!

[@Jackmackrel]

oh great, now i have to buy a tr 909 again

[@atommachine]

I used to run my 909 from the rhythm trk on a mc500 it sounded great, Mc500 had flam I think. I use a Tr09 and love it, It's output distorts very well through old desks and has a Dark 909 feel. Looking forward to your test.

[@silicon1138]

How are they connected? Are they all using standard midi or are the new machines using USB? I had loads of Trouble with midi jitter. The more I analyzed it under the microscope, the most I saw midi synths and groove boxes were all falling on random spots, not bang on. I ended up buying the expert sleepers USAMO which creates an audio sync pulse for a sample accurate clock and the DAW locks to that. Improved things a lot. Jitter is a real rabbit hole.

[@StarskyCarr]

In this I’m using them all independently. I set them all to the same tempo (they don’t all play exactly the same tempo). I then retimed all the samples so the first beat of the second bar was exactly in time to the sample.

[@kamilpakatheinfluence]

I have tried to reproduce tr-909 on Ableton for years and failed. It's those slight dieferences with timing and amplitude variances that make the origininal 909 have it's own character. It's hard to switch away from it. The feel is just right, when listening to the original.

[@wackerburg]

That's why I finally went with the original machine and never looked back. In the end it's something like "If you want X, get X - not Y."

[@cabeuscalling]

Great observation! I will definitely experiment with timings on my next project 👌

[@Bananskuden]

Now you got to try it with at least two more 909's.

[@loksync]

Internal clock is tighter than clocking the 909 externally apparently (214 samples of jitter from the internal clock vs 374 samples of jitter clocked from external midi clock (from the Innerclock Litmus test results)). Also, doesn't the firmware version (eprom) really matter with the 909? v4 firmware meant to be better midi. That's as far as my research goes from decades of wanting one but never getting one.

[@PacouJustPacou]

there is a v5 kit now from a DIY guy in france it fixes the MIDI issues

[@crunchysteve]

The old 909 probably didn't have anywhere near the speed of modern microcontrollers, it's possibly latency. Makes me think it'd be cool to try randomising latencies on a MIDI repeater using an arduino. Add a millisencond here, 2 or 3 there. Clock a 328p chip at 8Mhz instead of its usual 16Mhz, turn off the Thru method, do MIDI.write(MIDI.read) (not the correct syntax of the top of my head, but you know, more lines of code.

[@JayKaufman]

MIDI jitter on a device as old as the TR-909 is hardly surprising. Personally, unless I want my drums to be sloppy, I'd much rather they hit in time and then I can adjust accordingly in post production. That said, especially in hip hop circles the MPC-60 is revered for its MIDI jitter and shoddy timing because it adds groove to your drums.

[@DJ_312]

Intriguing...I used my TR-707 and recorded MIDI into Ableton Live 12. I do experience the same sloppiness with the recorded MIDI notes.

[@michaelpierce3264]

I have an original cr8000 that would swing like crazy.Ended up changing the plug on the worker supply and now it’s stiff robotic

[@mufakkas9731]

that was a big thing in the 80/90s, midi interfaces, Atari etc

[@mrrothers3024]

The Casio rz1 BPM is all over the shop when it's a master , I had my tr8 syncd to it and the BPM on the tr8 was drifting back and forth

[@davek4605]

@@mrrothers3024 yes I’ve noticed this. The timing of hits a bit off too even when slaved.

[@banksideleopards2791]

You definetly touch the point with this video, what gives this indefine feeling with 909 or 808 it's the groove not the sound, most of all clones are really good for this

[@Bigger-Circuitry-Bigger-SOUND]

Starsky what gives those machines “that thing” is the circuitry inside don't go further than that, how pleasing the (circuit topology/components in case of analog and converters/resolution in case of digital vintage) create the harmonic distortion/second harmonic/extra harmonics or as called in popular language “warmth/sound depth”! for example din sync sound exactly like the 909 because of the circuitry replica 1 to 1, people look everything in a synth except the most important thing that generates the sound The Circuitry

[@unclemick-synths]

People have become too hooked onto the grid and stress about everything lining up visually even though our ears don't see too well. "Look at this MIDI latency!" 😱 Quantize really is a groove killer. Rick Beato had fun quantizing John Bonham stems to prove it.

[@pizzagogo6151]

Thanks Starsky a bit of a fun discovery for the worlds greatest drum machine 😄I'll just point out this came up on KZbin as 6:06...? Im hoping this is edited & the "super special version" on your patreon etc is 3min, 3secs longer😮😅

[@SpikesStudio3]

Good spot. Had a giggle at that. Sorry best drum machine has to be machinedrum. . . Or maaaybe 808.💣

[@wackerburg]

The way I look at it; 6:06 upside down resembles the might 909

[@nnervecenter]

Just recently been able to play vinyl again. Listening to some Terrence Dixon this morning and thought the (great sounding) sloppiness in timing was down to the Soundburger struggling to turn the record

[@universalvibe72]

Starsky: your 909 is broken. Tru story.

[@yuho865]

Using the 909 as a master clock is/was pretty common for the feel of it, also it's easier to sync stuff to its crappy clock than the other way around

[@synthmaven]

Wonder if the RE-909 and the Steda have the same properties..

[@paulc7798]

Are all the kicks off as well? I remember ATFC saying that he uses a particular swing from Guru on his kicks to make them not bang on.

[@StarskyCarr]

its the same for any of the sounds. Probably the CPU doing its best but not having the power to do things bang on the beat. It complements my 'loose' playing style at least :)

[@slavestudios47]

i have a TR909 previously owned by Space DJZ & it sounds great BUT.. when i used it years ago with a live set that involved Ableton loops, lock groove 12s & other sequenced samplers the thing drifts all over the shop. alone, it gives a great 'human' feel, but when trying to clock it with other gear for a live show using DAW & vinyl, it would make a mess.. the TR-09 isn't as bad, but the og 909 a timing mess. we ended up using an SR16 to trigger the 909 it was that messy. looking back i now know why some techno records were so messy to mix. its a drift nightmare. and the more going on in a pattern, the worse it gets. now i know how Mills et al had those old Detroit trax sounding so groovy.. and a bollox to mix lol something to do with internal memory lag afaik. spent a lot of time reading up on it & found out all sorts about variations in quality of components & production issues. never meet your heroes innit 🙃

[@StarskyCarr]

Those older tracks are pretty sparse, maybe the dodgy timing help fill the gaps 😆

[@squarelanguage]

Curious if anyone knows how many milliseconds it's ahead of the beat, and is it the kick and hats that get pushed ahead or just the kick?

[@MusicoElectronico]

I use vintage drum machines in all my production work and that sloppiness is quite common... there's no prefect timing in vintage drum machines that's the charm of vintage gear.

[@ziggyzaagsma473]

Sure hear the moving around of the samples by quantizing them sounds phasing/glitching/aliasing

[@Nijnvenijn]

When I had a TR-8 there was a setting in there to either emulate the timing of the originals or to have everything locked in. I think you are on to something. If I use a sampler for drums everything tends to sound samey and unexciting.

[@richgsimisker6286]

Interesting! Maybe coming in early means it's not getting as affected by frequency masking from quantized bass parts.

[@StarskyCarr]

That would definitely work, but I’m not sure it was intentional.

[@richgsimisker6286]

@@StarskyCarr No, you're right - I'd put good money on it not being intentional! But maybe that's one of the factors as to why proper 909s have become coveted over modern recreations, where tightly-timed kicks get masked by equally rigid basslines, because they're too good technically. Just a thought!

[@StarskyCarr]

💯

[@scottbradley9164]

I purchased the midi files and found them very interesting. One thing they didn't tell me which I'd like to know is if different instruments (e.g OH/RS) hitting on the same step would have the same sloppiness?

[@StarskyCarr]

I’ve just had a look. Each sound has a similar amount of sloppiness but not identical on each hit. So the snare is as sloppy but not identical. With the hat and snare together for example they will both be off by a different amount. I may make new files that are much longer so you can select different bars for each instrument so you get the same 909 inconsistencies but not the same for each instrument). You should be able to DM me from the Patreon account.

[@NicholasBryantBonzaiSequoias]

It could be down to the 909 having a comparatively slow CPU (by modern standards..) and therefore some timing issues occuring at certain tempos (the calculating of tempo being constrained by the CPU clock rate). Obviously Roland would have made sure it was useable, but back in the 80's we just weren't used to 480 ppqn..

[@gixerags750]

I wonder whether it is from the age of the component's in the TR909 ?. Also do run an ErmMulticlock /Midronome etc ? Going to try this with my RD9, MC505 and MC909.. GREAT TIPS 😊

[@RaquelFoster]

Something I'd never heard before ... Johnny Morgan Synth Dreams (AWESOME channel) has a 909 video he put up 3 weeks ago, and it says the 909 has phasing between the snare & clap since they share the same circuit. I'm curious if the TR-09 or the Roland Cloud version do that. I don't imagine the ACB recreations in the TR-8S would take that into account or automatically tweak the timing just because you loaded a 909 kit. These don't seem like things you'd ever recreate unless you were using an actual full emulation of a 909, not just a sample or even an ACB version.

[@StarskyCarr]

Yeah that’s true. I remember someone talking about that when the TR09 was released.

[@wackerburg]

That phasing behaviour is also "dynamic", which makes it even harder to recreate in a DAW or with an MPC imho. What I mean by that is it depends on the Accents and volume relations.

[@BlackMan614]

I seem to remember someone analyzing the timing one of the original MPC models - known for the groove - and it wasn't "in the pocket" on the beat, but *consistently* off the beat (late) - but consistent up to just a few milliseconds. It was the slop (the Roland R-8 is notorious for it) that sounds bad.

[@katchosan8858]

What about Roland TR-8, there is option to set as TR 808 sequencer timing. Can you confirm, its doin a trick?

[@Shred_The_Weapon]

So, the samplers and emulations have better quantization than the original 909?

[@fritsvanzanten3573]

Coincidentally (or maybe because an article on a site like Music Radar) a few weeks ago I looked up an scientific article about perception of small deviations of perfect beat. Even more coincidental today I worked on an unsteady beat (old CD-track) in my DAW, sending an email to a musician friend about the attraction of these small deviations. Are they sloppiness or are they the effect of emotion in playing (and yes, the players starting hurrying with the chorus in sight, and slowed down after the climax)? Maybe perfect beat kills something in our appreciation. Maybe perfection is unnatural.

[@markszili2577]

the 909 kits on the TR-8s has the same variations I'm guessing it's by design to give it that authentic feel

[@StarskyCarr]

I wouldn’t have expected that. I bring out the TR8 in my main demo - just as an alternative but don’t play it.

[@brandonm8385]

This is the internal jitter.

[@livewarp]

35 years old it could be down to the aged electronic components. Probably all 909’s are different anyway?

[@MP-fs4ey]

I reckon it's the same with the TB-303 compared to all the clones........some special magic the original had with aged circuits etc that can't be replicated with modern 'exact' tech

[@Nightingale1887]

If the clock generator is analog, then the offset might be due to component value drift due to age? Hard to compare now because all 909's are vintage at this point. Component spec is also a thing with the tolerances being tighter now than back in the day. Like every 909 has a different sound to it. Filtering, cap values drifting etc.

[@StarskyCarr]

True. I caveat the whole of my tr909 comparison demo with just that. Taking wide tolerances, degradation, ambient conditions over 4 decades combined with a number of different hardware revisions leaves us with anything f it a definitive ‘ideal’ tone.

[@damonbennett5911]

Is this strictly an internal clock phenomenon. Does it do the same thing when synced with MIDI Clock?

[@sqwook]

so it is grim up north, how else would you find the time

[@StarskyCarr]

The grimness of the North spawned the Industrial Revolution and the dawn of the modern age. In the Mediterranean folks sat with family eating great food and drinking fabulous wines with family, watching the sunset over the Dolomites. Here we spent the evenings feckling in sheds, making weird machines drive by coal to avoid speaking to our wives or interacting with kids. And so it continues … nothing to be ashamed of 🤣

[@sqwook]

@@StarskyCarr starsk i was referencing the banger "its grim up north" by the JAMS which has to be one of the most beautiful homages the sounds of that industrial revolution, also the shit weather we are likely both experiencing as I'm just over in Dublin (which is pretty far north on the best of days) and i think your near liverpool. Wouldn't mind some of that Mediterranean weather right about now

[@StarskyCarr]

@@sqwookhaha I got that. But it’s a theory I’ve held for years. Grimness bears fruit. We all need a light and ours has the magical ability to become the giver of light to the masses. Let there be electro. Let there be techno. Let there be light 😂❤ I can imagine a Victorian northerner in his shed back in the day saying something similar about pumping water from coal mines and keeping the mills running with his miraculous invention. It’s the Posh pOSH travelling life for me 🤣

[@scratchanitch]

Another geek idea: try using the tape sync to see if it has any impact on timing?

[@StarskyCarr]

Aaargh … no more 🤣🤯

[@johanneschristopherstahle3395]

Espen Kraft did talk about this. I think it was in his episode about the 626 (or maybe the 505). That's why he prefers old drum machines over modern sample players. Those imperfections are what makes those units perfect. Interesting to see that even the MIDI data is not on point. Does it match up with the audio or does it have it's own timing?

[@StarskyCarr]

The MIDI is taken from the audio, using Logic's flextime to create MIDI trigger points.

[@johanneschristopherstahle3395]

Ah, thanks for the quick response 😊

[@tinplategeektoo]

The key question we need to ask is : if you took a second TR909, would the results be identical? If the misalign is deliberate then yes but if it is down to the percentage tolerance differences in component values then you may see differences.

[@StarskyCarr]

There’s always more funding needed for more research!

[@alphanumeric1529]

Lol, i struggle so much with my volca sample, that thing is loose as hell. Drives me crazy as I'm trying to stack it with other devices to create complete sounds, but the Sample is always ahead, behind, all over the place. Frustrating... so much so I haven't been using it. BUT, maybe I should... just use it for individual drum hits, and roll with the slop? I'll give it a shot.

[@BxCx666]

Are all of the tracks on 909 has the same “swing”, or it is slightly different? Anyway I suppose the variation in timing should make sounds more present if they’re played along with perfectly quantized track (e.g. bassline).

[@Heathcliff_hensel]

That is vetry interesting good catchs

[@cryptout]

This is known in the 909 community, if you program in all instruments it becomes even more "sloppy" or "groovy" depending on how you look at it.

[@JimmyHandtrixx]

yeah maybe....but nobody notice this BS on a track! i have had 2 909 and now have rd9. its the same. such crap.

[@pablowentscobar]

Interesting. Gives it a more human/IRL feel. Either no, or very few humans are going to have perfect timing. My question is, did Roland do the on purpose Or was it a happy accident. And, if they did, why didn't any of the clones try to recreate it? I would have to imagine the companies that made clones realized this. Good stuff Starsky.

[@klinkske]

Imagine using the TR 909 trig out to play a synth. That gives it even a bit more 'feel' - try it? :) Or use it with the midi out to play notes. (I wonder how few people do that)

[@StarskyCarr]

Nice idea

[@a1o1]

Just missing the RE-909 to add to the comparison matrix

[@StarskyCarr]

I’m working on it! My mate has one but has yet to finish it. Another has one on the other side of the world!

[@tehedx]

It appears that it's due to the TR-909 having a single CPU that needs to do it all. So when it had to hit the beat, it could be processing some other task and had to finish it first. Music in old games wasn't always in time either for the same reason.

[@jemski]

The original 909 used shuffle which is technically different to swing. None of the later versions used shuffle...

[@StarskyCarr]

What was the difference?

[@terminalbliss]

I think the reality is that by having a little push and pull it will also have impacts on phase in different parts of the track (e g. Kick will be out of phase with bass and impact volume at points) which over the course of a song will yield to yet more variations. The brain is good at pattern recognition so the more variation, even if minor, may subliminally hold our attention more. It may also show how phase aligning every bass and kick hit in chasing perfection may not be the right thing to do from a "keep it interesting" perspective. I stopped 100% quantize a while ago as i really dont like the feel, so this isnt a surprise. You dont need a real 909 just dont 100% quanitize and leave some variation as it will keep our attention better.

[@StarskyCarr]

Absolutely.

[@falangistavaleroso9689]

So you did a Logic swing template out of 909 internal sequencer? You late, I already did in the late 90s... lol, this is something we techno guys know since the late 90´s.

[@StarskyCarr]

Ooh… me too. I didn’t have a 909 back then so used samples from vinyl to do the same thing.

[@falangistavaleroso9689]

@@StarskyCarr I paid 240 euros for mine in 1994, it was from a beach orquesta from Benidorm, they thought I was crazy by paying as much for it... now 6000e!

[@lukassbeataddicts]

That make me wonder if Roland Cloud plugin has this "sloppiness" perhaps not.

[@mu_zines]

Nope, it's bang on - which probably suggests Roland think the wonky timing on the hardware (if this is a general thing, not unique to Skarsky's unit) is a bug, not a feature...

[@prodigalretrod]

So I take it that whatever level of sloppiness is present, it is the same across all drum tracks?

[@StarskyCarr]

I’ve not tested that thoroughly. But on the complete beat there seemed to be similar discrepancies.

[@3dsman]

Probably the older equipment's bit depth limitations are causing this. Consider if hypothetically an 8 bit CPU were given a 256bpm rhythm pattern (ridiculous I realize but stick with me)... if the CPU were asked to 'swing' the tempo it would be pretty successful since in an 8 bit world a 256bpm (or 16, 32, 64bpm) would result in easily divided results. Give it 113bpm and things are fine with no swing but once you start swinging the tempo the bpm probably doesn't result in perfectly divided pie slices and the 'roundoff' errors give you some pie slices bigger than others and results in the 'organic'-ness everyone raves about. Also consider that even if you think my theory is bs... sorry... poppycock... that if you have layers of these 8 bit limitations including the source samples and the clock/cpu frequency and all the other algorithms that are simultaneously taxing the relatively primitive silicon of the day... errrrrrorrrrs will be rampant. Compare that with today's high CPU speeds, huge bus widths, and 32 of 64 bit cpu's (compared to old 8 bit)... and modern drum/rhythm machines probably 'NAIL' the calculations so perfectly it sounds robotic. Possibly even the limited RAM of yesteryear compared to today's equipment also results in compromises that today's equipment doesn't face. One person's theory.

[@pheotus]

Prefer the sloppy. ❤ I think people spend too much time quantizing everything and trying to make it perfect. If it sounds good, then it sounds good. Leave it alone not everything has to be syncopated.

[@christianthompson1473]

So all those years ago when people were saying the timing was better on the TR.... they actually meant... the timing on the tr was worse. 😬

[@christianthompson1473]

*but in a good way (sometimes)

[@electriceric7472]

Did you record multiple takes at each setting though? I feel like your regular homegrown midi jitter will give you this, but apologies if you accounted for it.

[@StarskyCarr]

I recorded 16 bars of each. It could be midi jitter - but it seems too loose. I think it’s more likely to be CPU limitations.

[@abwilson2852]

Jitter doesn’t sound good whether it’s coming from a 909 or a sloppy drummer. Human feel is about landing off the beat consistently in the same place every bar, that’s what creates the groove. The jitter from a 909 is inconsistent. The beats land in different places every bar. I’m saying this as someone who’s owned a 909 for nearly 30 years. It’s great fun to program, but if I want to record it I’ll trigger it over midi.

[@christothegreat1]

What about the timing on the Roland Cloud TR-909 software drum machine?

[@StarskyCarr]

I think that’s bang on

[@mu_zines]

@@StarskyCarr It is, just checked...

[@TeslaDanser]

Did u get the idea from Distort the pre-amp?

[@StarskyCarr]

Don’t know what that is… I got the idea from buying a tr909 last week and having a play.

[@naoarte808]

I wonder if the age of the device is a factor?

[@StarskyCarr]

Of course it will be. But to what extent and changing what characteristics is a whole universe of unknowns.

[@naoarte808]

@@StarskyCarr That's everything I love about this gear! 😀

[@paulc7798]

You need to also bear in mind that Logic is not the best for waveform display and transient detection. Each hit of the 909 is different so Logic will be different. Have you tried the same test on another DAW?

[@StarskyCarr]

What does that mean? Logic is displaying the waveform which would be the same in any audio editor. Each hit does look different. I’m not sure what you’re getting at. Are you suggesting that Logic doesn’t display the waveforms accurately? It’s the first time I’ve heard that. Even so, surely it displays the amplitude or activity at least at specific points in time? which is what I’m discussing. I have to say it’s hard to believe it can’t display waveforms accurately, when basic software has been able to do this for decades. I’m not saying you’re wrong btw. Just that this is news to me. I would’ve expected a leading audio editor to be able to represent waveforms at least as accurately as stuff I used 30 years ago… but you never know!

[@mu_zines]

@@StarskyCarr Logic's waveform drawing is varying levels of problematic, yes - but I don't think that will affect your findings in this case...

[@jimifiTube]

Just out of psychoacoustic curiosity, what was the range (in ms) of variation? In other words, how many ms away from the absolute 1/8 note's start times were the real 909 you recorded? And out of engineering curiosity, are the 909 timing imperfections absolutel or relative to tempo?

[@StarskyCarr]

You can probably work it out from the MIDI resolution. Between 234-240 when 240 is a division. From the time code it looked like IRO 4ms - each second has 25 frames etc so it would take sitting down with a calculator.

[@jimifiTube]

@@StarskyCarr That makes sense. I can reliably indentify the quantized & nonquantized 8 bar loops with the full beat, but recreating and soloing just the HH I could not hear/feel a difference. I suspect I am hearing (and preferring) the novelty of the spectral variations in the authentic/imperfect 909 and not the temporal Russian Dragon-iness of it all.

[@anreoil]

But humanizing is what we add in DAW anyway, it's a standard feature. Otherwise all music will sound the same. It does.

[@StarskyCarr]

I wonder how many people actually use it.

[@SEFFOfficial]

Imagine everything being completely on the beat, would be dead bloody boring!! 😅

[@adamholmes3135]

Maybe Eric Kretz programed them. 😅 He is all over the place, in a really good way.

[@annother3350]

Did you record using the internal sequencer to drive the drum machine or midi hits? I used to have a Drumtraks and the midi was very poor but the internal sequencer tight

[@StarskyCarr]

The internal sequencer