Elektron-Users

rgmccaig wrote:
Ok, I'll get together a bunch of tracks, some sample accurate some not. There will be no overlap though, I won't post the same track twice (one entirely sample accurate, one not) because it bares no relevence to a real world scenario.



No, that method is still a little flawed.

A complete stranger hands you a track he made on the street and asks you if its sample accurate or not. Can you answer this accurately? Lets find out.

Um no, how is that the reality? That would never happen in reality. The reality is actually: will people enjoy my track better if I can make it closer-to-sample-accurate. _That_ is the point.


It's pointless to test _different_ tracks, then you will be testing mainly people's preference for different compositions. This is known as a 'confound'; you must 'control' for this. You will not be able to balance out that confound unless you use a ton of different tracks and different subjects (and find a way to filter out your _own_ bias ie. choosing tracks to represent tightness that are 'better' in other ways) - i don't see that happening for this experiment.

The question we are interested in is, given the track i'm working on, will that particular track sound _noticably better_ if the timing is made tighter. Best way is two use 2 versions of the same track, then the experiment is balanced.

It's theoretically fine if you want to do a 'between subject' design: use 2 versions of each track; but for each
subject, only give them one version. Then average the numbers between subjects to find if there's a trend towards liking the tight versions better.

Problem is, you still need a lot of subjects that way, in order to cut through the randomness associated with using different subjects.

PS- in order to test the population of interest, imho you should use subjects that actually like/listen to/dance to electronic music; also you should test it on a decent 'body-rocking' system so that people can Feel the groove.

Argh, no. The question and discussion was definately based on psyhcoacoustics to start with, like you said "adjusting a knob and hearing a difference". But anyway, you say my testing method is flawed because it simply uses compositional preference, yet you say that the test is done to see if people enjoy one way more than the other? Your method is based on exactly the same compositional preference that you suggest is what flaws my method. For timing to be identified as the reason for this preference, we must first be able to identify this timing... pretty straight forward. If people can't identify timing as sample accurate or not in single clips, how much of an increase in enjoyment do you think people would gain for sample accurate timing? I am happy to do both, yet the test you suggest cannot be done on this forum without bias.

ZiggY wrote:

Argh, no. The question and discussion was definately based on psyhcoacoustics to start with, like you said "adjusting a knob and hearing a difference". But anyway, you say my testing method is flawed because it simply uses compositional preference, yet you say that the test is done to see if people enjoy one way more than the other? Your method is based on exactly the same compositional preference that you suggest is what flaws my method. For timing to be identified as the reason for this preference, we must first be able to identify this timing... pretty straight forward. If people can't identify timing as sample accurate or not in single clips, how much of an increase in enjoyment do you think people would gain for sample accurate timing? I am happy to do both, yet the test you suggest cannot be done on this forum without bias.

Ziggy - the wheel fell off your barrow a good few pages back. You are welcome to keep posting of course but I do think your time would be better served getting the wheel back on straight and then come back with something positive to contribute. What you are offering is only blind repetition based on a deep misconception coupled with anger fueled by your own inability to see reason.

David.

innerclock wrote:

ZiggY wrote:
I didn't say which tracks in the clip were sample accurate, why? because people who listen to your music have no idea what track is sample accurate or not, they have no idea what instruments were used to create what.

And again Ziggy - sample accuracy is not important on this thread outside of testing errors and deviations - what is important in listening/real-world examples is TEMPO/STEP/DURATION/EVENT precision. The people that get this know already. You keep flipping the same old hocus pocus about hearing timing drift in a complete mix. It's a crap argument dude.

It's bloody simple - If I take two drum machines and sync them up. One machine playing hard-quantized 8th Kick/Snr pattern - the other playing hard quantised 16th hats OK - now, if I love the feel/groove of the first machine playing solo (no hats) right - now I push up the fader with the 16th hats in sync.

OK - they are in sync - Fine. BUT - they sound wobbly against the Kick/Snr and take away from the original feel.

If I pull down the 16th Hats volume - the tight feel is there as before.

Now - if I record these two machines - the Kick/Snr to [L] and the 16th Hats [R] while in sync to SF8 as I have done many times and zoom in on the file - what do I find?

Very simple - the Event/space durations on the Kick/Snr pattern are very precise - no deviation more than 5 samples at 44.1 kHz across the whole 2 minute recording.

The Hi Hat 16ths in contrast move around by up to 128 samples from where they should fall.

If that were two drummers playing on stage - any member of the audience with a musical/rhythmic ear would tell you the guy playing Hats is not a patch on the guy doing Kick Snare.

This is the whole focus of this thread - the SPS-1 is the guy playing hats in this example and the rest of my band want him to take some lessons..... :-P


David.


No, you guys are quickly progressing away from your original statements that you can hear and notice these deviations and could pick them in blind tests because, what did you say again? your hearing is perfect? :roll:

I know exactly what this thread is about, we have been over it many times but you keep drawing a blank it appears, I use sample accuracy as a reference for comparison exactly like you did. This thread has moved so far away from what NEEDS to be done to simply what people prefer it appears. You cannot get more in sync than it being sample accurate, you are suggesting that people prefer music that is more in sync. Regardless of whatever tests you do in your studio, for any of it to be relevant you must prove that people notice it and get more enjoyment out of a piece of music proportionally to how accurately in sync it is.

And again, for any of this to be a problem you should be able to pick it as the problem or as a track not entirely in sync just by listening to it.

ZiggY wrote:

No, you guys are quickly progressing away from your original statements that you can hear and notice these deviations and could pick them in blind tests because, what did you say again? your hearing is perfect? :roll:

I know exactly what this thread is about, we have been over it many times but you keep drawing a blank it appears, I use sample accuracy as a reference for comparison exactly like you did. This thread has moved so far away from what NEEDS to be done to simply what people prefer it appears. You cannot get more in sync than it being sample accurate, you are suggesting that people prefer music that is more in sync. Regardless of whatever tests you do in your studio, for any of it to be relevant you must prove that people notice it and get more enjoyment out of a piece of music proportionally to how accurately in sync it is.

And again, for any of this to be a problem you should be able to pick it as the problem or as a track not entirely in sync just by listening to it.

Wrong again Ziggy - that example I just gave IS THE VERY BASIS OF THIS ENTIRE THREAD - IT IS THE FOUNDATION OF MY WHOLE DEBATE - I HAVE NOT SHIFTED MY POINT OF VIEW by a single solitary sample. There, in that example is my issue with the SPS-1 in a nutshell - plain and simple to understand and test by anyone. It still stands.

David

ZiggY wrote:

And again, for any of this to be a problem you should be able to pick it as the problem or as a track not entirely in sync just by listening to it.

I can't believe you dont get it. The example I gave is sound - two machines - one in time - the other sloppy.

If I'm in the studio right, to record a song - I get a session drummer in - he plays tight - nails the take perfectly. OK next day - I book a conga player - he lays down 15 takes over a whole day but he doesn't cut it next to the drummer - sloppy and loose. Ok - by your standards now - thats just music - live with it and move on right?

Wrong - I book another persussionist, and another, and another if I need - until I find the one that has ears AND feel and locks it down next to the kit.


He gets paid - the others don't get anymore work.

Now I book the bass player..... and so on.

This is how music is made. For real. Not fantasy.

Get on the web and have a look at how many great session musicians Fagan/Becker/Steely Dan went through just to get the feel right that they heard in their heads during sessions. Every one of those players would have cost a fortune to book and every take was probably gold in it's own right - but they wanted a very specific feel and quality - and Fagan was/still is a timing fanatic - that doesn't make him nuts - he just wanted it to feel very specific and he knew when it was 'on' and when it was 'off'. Do you think anyone told them at the time - 'Come on guys - that bass part is just fine - no one will notice in the final mix?'

Do you know for creating demos on his solo work he used to use an MPC-60II for drums which is more than tight enough to begin with but I'll tell you a little secret from Roger Nichols (Fagan/Steely Dan Engineer) - to get his demo drum feels right while he was composing - he had a rack of 8 x Roland SDE-3000 Delay Lines strapped over every output of the MPC-60II. Why?

He had them all set to 50 ms so he could push and pull individual instruments by as little as 1ms to get the feel he wanted but still driven by the tightest drum machine available at the time at any price.

You don't have to put up with slop - human or machine - especially if you can hear it and you can fix it.

Come on man - wake up.

Ziggy, I see that the issue may not be very important to you. I could not hear anything wrong with your sample, I then checked in a waveform editor quickly and could not find anything wrong. But also, even if the ear can be very precise to detect timing variation, it does so when two events are happening at the same time or not. Detecting variations on the order of 1 ms on a same repeating metronome like pattern is much much much harder (I think on this order it is next to impossible).

I do know from practicing a lot with a metronome that after being in the zone for a time, you can start to pick up very very minute variations in groove and timing (I have this one exercise when I pretty much know when playing a note that I am correct or lagging by 2ms, and in this case it is possible for me to shift my groove around by a mere matter of ms). That doesn't mean I can play any better than a MD, or in fact that anyone can. But really this kind of tightness is not the one that is discussed here.

On the magnitude of ms or less, you are starting to enter sample-realm. And what we can do with the MD is not something you'd attempt with a real drummer, but things that are possible with DSP and sample-like accuracy. For example putting two bass sounds together, or using their interlocking phase effects to do weird and beautiful sounds, or whatever. And that is something where everybody can hear ms wise differences. For example:

http://bl0rg.net/~manuel/md-slop.mp3 (best heard on speakers and not on earphones)

The first is bd samples i lined up by hand in the audio editor. FIrst 1 ms difference, then 2ms, then 3, then 5ms. Can you pick out which kick has the difference between left and right? I know I definitely can, and most people (not into music producing) I asked definitely could too.

Then after that is a recording of the machinedrum, played against a loop recorded immediately after. Again right and left channel. Does that sound to you like a little phase-shifting melody is played? To me it does, and that is definitely not what I was looking for.

So, I know these examples are construed, but they for one show that you can definitely easily hear 1ms timing differences. And the other side of hearing such things is the pretty unconscious things of "tight machine robot impossible to achieve" groove versus "yeah tight drummer" versus "beginner" groove feelings.

What I don't understand is your need to go through this again and again. I mean, ok, you can't hear it, it doesn't interest, why do you even care about this thread?

daswesen wrote:
Ziggy, I see that the issue may not be very important to you. I could not hear anything wrong with your sample, I then checked in a waveform editor quickly and could not find anything wrong. But also, even if the ear can be very precise to detect timing variation, it does so when two events are happening at the same time or not. Detecting variations on the order of 1 ms on a same repeating metronome like pattern is much much much harder (I think on this order it is next to impossible).

I do know from practicing a lot with a metronome that after being in the zone for a time, you can start to pick up very very minute variations in groove and timing (I have this one exercise when I pretty much know when playing a note that I am correct or lagging by 2ms, and in this case it is possible for me to shift my groove around by a mere matter of ms). That doesn't mean I can play any better than a MD, or in fact that anyone can. But really this kind of tightness is not the one that is discussed here.

On the magnitude of ms or less, you are starting to enter sample-realm. And what we can do with the MD is not something you'd attempt with a real drummer, but things that are possible with DSP and sample-like accuracy. For example putting two bass sounds together, or using their interlocking phase effects to do weird and beautiful sounds, or whatever. And that is something where everybody can hear ms wise differences. For example:

http://bl0rg.net/~manuel/md-slop.mp3 (best heard on speakers and not on earphones)

The first is bd samples i lined up by hand in the audio editor. FIrst 1 ms difference, then 2ms, then 3, then 5ms. Can you pick out which kick has the difference between left and right? I know I definitely can, and most people (not into music producing) I asked definitely could too.

Then after that is a recording of the machinedrum, played against a loop recorded immediately after. Again right and left channel. Does that sound to you like a little phase-shifting melody is played? To me it does, and that is definitely not what I was looking for.

So, I know these examples are construed, but they for one show that you can definitely easily hear 1ms timing differences. And the other side of hearing such things is the pretty unconscious things of "tight machine robot impossible to achieve" groove versus "yeah tight drummer" versus "beginner" groove feelings.

What I don't understand is your need to go through this again and again. I mean, ok, you can't hear it, it doesn't interest, why do you even care about this thread?

you've listened to the clip and could not hear anything? It contains all the things you've listed, including layered tracks that develop a big sloppy shift. The difference between my clip and yours are that mine aren't construed outside the context of a piece of music. Of course, david likes to say I don't get it and he can use all the metaphors he likes but he still hasn't identified the location in my clip where is suddenly falls apart timewise. Why? Why can't you hear it? Why can't he hear it?

Why am I still here? Look at some of the posts people have done in this thread. Some of them even arrogantly attacked the guys at elektron... yet so far it is only over a very construed studio process that nobody has even been able to link to having an effect on a final piece of music. I'll put it very clearly, if someone can identify when the wheels fall off the clip I have posted (which is entirely within the context and examples of "problems" that people have posted in this thread as the very reason for the MD needing improvement) then I will send an email to elektron myself and say the MD is broken.

innerclock wrote:

ZiggY wrote:

And again, for any of this to be a problem you should be able to pick it as the problem or as a track not entirely in sync just by listening to it.

I can't believe you dont get it. The example I gave is sound - two machines - one in time - the other sloppy.

If I'm in the studio right, to record a song - I get a session drummer in - he plays tight - nails the take perfectly. OK next day - I book a conga player - he lays down 15 takes over a whole day but he doesn't cut it next to the drummer - sloppy and loose. Ok - by your standards now - thats just music - live with it and move on right?

Wrong - I book another persussionist, and another, and another if I need - until I find the one that has ears AND feel and locks it down next to the kit.


He gets paid - the others don't get anymore work.

Now I book the bass player..... and so on.

This is how music is made. For real. Not fantasy.

Get on the web and have a look at how many great session musicians Fagan/Becker/Steely Dan went through just to get the feel right that they heard in their heads during sessions. Every one of those players would have cost a fortune to book and every take was probably gold in it's own right - but they wanted a very specific feel and quality - and Fagan was/still is a timing fanatic - that doesn't make him nuts - he just wanted it to feel very specific and he knew when it was 'on' and when it was 'off'. Do you think anyone told them at the time - 'Come on guys - that bass part is just fine - no one will notice in the final mix?'

Do you know for creating demos on his solo work he used to use an MPC-60II for drums which is more than tight enough to begin with but I'll tell you a little secret from Roger Nichols (Fagan/Steely Dan Engineer) - to get his demo drum feels right while he was composing - he had a rack of 8 x Roland SDE-3000 Delay Lines strapped over every output of the MPC-60II. Why?

He had them all set to 50 ms so he could push and pull individual instruments by as little as 1ms to get the feel he wanted but still driven by the tightest drum machine available at the time at any price.

You don't have to put up with slop - human or machine - especially if you can hear it and you can fix it.

Come on man - wake up.


A session player plays tight? How tight? Have you every measured the tightness of a session musician? Do you think a session musician plays it indentically in every take? I need to walk up? Have you actually read what you are posting?

ZiggY wrote:
.. yet so far it is only over a very construed studio process that nobody has even been able to link to having an effect on a final piece of music.

Have a read of my comments about Fagan/Steely Dan - the process I have described are not abstact at all - it is how music was and is still made by profesionals everywhere and it has every effect on the final outcome in the music.

You need to get outside your box and see whats out there my friend.

ZiggY wrote:
But anyway, you say my testing method is flawed because it simply uses compositional preference, yet you say that the test is done to see if people enjoy one way more than the other? Your method is based on exactly the same compositional preference that you suggest is what flaws my method. For timing to be identified as the reason for this preference, we must first be able to identify this timing... pretty straight forward.

No no, think about it again.

If people rate 'SAMPLE A' better than 'SAMPLE B' and the two samples are the _same song_ except for tightness, we have clearly shown that people prefer tightness, get it? That's what is meant by balancing.


Btw, don't assume that my stance on this topic is exactly the same as everyone on my 'side'; the way I see it, there are multiple valid questions to be asked here:

1. Can timing deviations of 2ms make an audible difference in a track? (clearly the answer is yes, for reasons explained by daswesen etc.)

2. Does that audible difference tend to sound 'better' when tighter? (this is subjective; but for innerclock personally we know the answer is 'yes'.. I suspect for a majority of musicians the answer would be 'yes'... that alone is reason to improve timing; we want to enjoy playing the instrument, right?)

3. Does 'joe average' in the audience prefer a song more if the timing is tighter? (A well designed study could answer this question, however it must be done properly.)

4. Does 'joe average' know that the reason for his preference is 'tighter timing', or does he just know that he likes it better? (I don't really see why it matters, but it could be theoretically interesting to know.)

5. Could 'joe average' identify exactly in which tracks, or when, the timing slop occurs? (This is a red herring as far as I'm concerned, I mean, who cares?)

6. Would better sps timing make it easier to cut/edit loops as part of the recording process? (Definitely yes, I would say.)

7. Would it improve the capabilities of SPS overall to give the option of closer-to-sample-accurate timing? (this issue depends on the other answers, and is complicated; personally, given the other answers, Id say it's a 'yes'.)


These are all valid questions;

except that debating question #1 is a waste of time, because anyone who works with electronic music (should) already know the answer.