page 1 | 38 posts new topic Store  
1 | 2 | 3
thread starter

I want to share with you guys, two opinions, seemingly dissagreeing.

One hexfix of the outfit Velvet Acid Christ, was interviewed for the Matrixsynth blog, and said: "
and it was the follwing that caught my attention:

link

"i spent years on the pc pulling my hair out, ever since cubase with audio in it, midi has sucked on the pc. cubase 2.8 was the last tight midi i ever got on a pc. i tried everything. and always got jitter of about 13 ms with even steinberg’s midi interface on external midi synths and drum machines.
It was an evil battle. i spent years trying to figure it out. no go. I came from the Atari st with notator and cubase. this was super tight and ultra responsive. nothing on the pc came close for midi. let’s not even get me started on how bad cubase sums and mixes digital signals. sonar 6 and saw studio kick the crap out of cubase for audio summing.

finally got a mac with an amt8 midi. running logic express.
WOW 3ms of jitter, just like cubase on the atari. it records me perfectly, something the pc could never do, especially when i was playing fast..
Logic is not much better at summing, but it records better, and i mix outboard now anyhow so. still saw and sonar 6 are king for digital summing and mixing.
But for midi, its logic all the way. it’s tight. cubase is not.

these are just my opinions. all based on the last 20 years of working on gear. It's all opinions. don't hate if you don't agree. it's like arguing that you like apples better than oranges. I'm not trying to tell anyone how to work or what to buy. I'm just talking about my own personal tastes and likes and dislikes."


Ofcourse, one could argue that in fact, he's comparing apples and oranges, and saying oranges taste sweeter.
I dont want to question his dislike for pc and favour for mac. Ive used the two of them, they both have their faults.

I am however curious about the question of midi timing.
A gentleman called kiwiburger, on the gearslutz site, said: "

"I was a loyal Ensonic sampler user for many years, and owned most models at some time. IMO, the D/A converters and/or analog stages were absolute shyte, and couldn't handle low bass. This was very evident on some samples that always clicked and farted on the low end. So there is plenty of distortion and noise going on to thicken up the sound. Not to mention the reverb that always defaulted to ON when you took a sample.

I use various software samplers, and have zero regrets. I've tried converting the old Ensonic samples, but basically my tastes have changed and I don't like them at all - not even for nostalgia.
There definately are some differences, when playing exactly the same sample - with no effects or filtering or anything.
Here is what is missing:

Noise - don't underestimate the effect that low level white noise has. I hate noise, but then again - sometimes when it's gone, I miss it. Analog dithering perhaps. (…) There is a an acoustic analogy - sound lacking bass can effectively have the low end restored if you have some low end noise to interact with the sound. Don't underestimate noise, whether in hardware samplers or analog summing or mag tape, etc.

Distortion - I have no doubt the cheap analog circuits added harmonics that thickened the low end.

Sample rate - (…)

Erratic Midi timing - the timing slop on midi hardware is shocking. I used to tear my hair out. Now I have perfect, sample-accurate timing - and it's too good. There is something about a little randomisation that our ears love. Whether is subtle wobbly tape, or erratic midi timing - a very, subtle, small amount of randomisation can be a very good thing."



Now I really dont want to get into a mac vs pc thing, and its not about hardware versus software either.

But there are still some myths surrounding digital music making that pop up now and then.
Keeping the channel faders up to 0dB (analog throwback), pci cards that are supposedly more prone to EM interference (soundblaster throwback) and then there's this midi timing thing.

Debunk please?

When have you run into sloppy midi timing; did it give a pleasent looseness to the music, or did you hear a digital mess?
Replies

I've heard, for years, people talk about hardware sequencers (in general) being tighter than software sequencers are (in general). I'll reiterate your pre-emptive bit about not wanting to get into hardware versus software or mac versus pc, though. I'm mainly chiming in to say that I've worked with hardware, software, mac, and pc and never found really noticable, inherent differences.

Is there any kind of "control group" way of comparing X to Y to Z in a way that would yield clear, statistical information?

Also: With piano rolls and other editing interfaces, it seems like playing and fixing the playing and just entering data directly into the sequencer (either step recording or using a mouse) exist as dark matter in the question of looseness vs digital mess.

Well, yeah, from this interview we have no way of knowing whether this was "MIDI Slop", or Cubase problems, or what have you.

You list some good myths, there. I wouldn't be suprised if some software, especially older software, has MIDI jitter.

Now, mind you, some people have bigger issues with their music than 10msec of timing.... puh leaze!

Still, back in the days when I actually listened to EBM like velvet acid christ, when I set up my two computers using the same software and set the bpm to the same number, they were off by 10%. 10%! And yet the bpm of Pro Tools gave you the option of a decimal place. HAHAHA. Thats not the same as jitter, I know, but it leads me to say "i wouldnt be suprised if THIS myth holds up.
After all, jitter could be the sum of errors caused by software, the oerpating system, the MIDI card, the MIDI standard, and the synth hardware.
Thats a lot of possible culprits. not counting the DAW that must be used to actually measure the jitter.

zach said: "
Is there any kind of "control group" way of comparing X to Y to Z in a way that would yield clear, statistical information?"


Perhaps the best would be to record the sound in a DAW and start counting samples/msec.
You'd have to assume that no jitter was introduced in the recording.

Unless you had an oscilloscope

SoundOnSound has a whole article on the various ways to combat MIDI Jitter
link
#1 seems to be killing background tasks.... this was back in 2001.

Oh man this is bringing back memories.... yes, I too used to use a command called "THIN MIDI CC DATA" so that i didn't try to cram too much down those little MIDI cables....I remember hiccups and blurps (not jitters, but little fits of bad midi timing).
I haven't sequenced 6 hardware synths at one time in ages.... kinda forget about the limits.

I also remember experiencing bad timing over long MIDI cables, i had to rearrange my studio and throw out the 20' MIDI cable.

zach - re: dark matter: you mean you can always go back to the editor to fix it?

ml - its indeed good to state a difference there. Jitter will never be a good thing.
Thinking about how sloppy playing will introduce feel into music (swing on sequencers?), I guess my concern is, can midi timing ever wreak the nasty havoc like digital timing (jitter).

10% is something!

As always, the pressures to have the best tools available to me are non existant for my amateur needs,
but its always good to know.
Especially with all the marketing hype.

true about the long cables.

midi over usb was another one discuscussion that seemed to never get solved.
the best I heard in the end was that, when usb started, audio companies did not yet write proper drivers.
usb1 should have been sufficient.

even a few msec of timing difference can be a major pain in the patookus when your Computer sequencer, and your drum machine, are both playing drum sounds. Nasty ass flanging, phase dropouts and flamming where there should be no flam.

13msec of MIDI jitter can suck ass when you are comparing two things.
And some people used to sequence 10-20 synths and drum machines all at once. if they were all 13msec off from each other, that could really suck nards.

13msec of jitter on a metronome is, well, not a big deal.

Still ,I suspect these two people you quote never heard of not using 100% Quantize on everything.
First I have a question:
If you mixdown your tracks to an export file, doesn't that alleviate any jitter in the usable recording?
I've only used Live and Cubase, and I always always hear more jitter in Live, especially as plugins in use start to accumulate. This is after all the standard DAW tweaks for PC, naturally. I haven't noticed bad jitter in Cubase since Cubase VST version 3. THAT was pathetic! Granted it was running on a Pentium II 333 Mhz, but still.

FWIW, I'd still use my Atari ST today for MIDI pattern sequencing if it didn't crash all the goddam time. The monochrome screen's kind of tiny too, but works nice if it's positioned higher than your head. It used to be perfect with the Akai S2800 display right below it.

what Ive heard about any time this is brought up is that cubase for the st was perfect.
so did the advent of audio in the sequencer mess things up?

there's different things that made the atari st so tight. the midi port was integrated as a serial port into the machine itself, and the OS was a monotask. So whenever something had to be sent, it would get sent real fast cause the overhead of putting something out was really low, and no other program would get in the way for your app to receive your bytes. This has changed quite dramatically, but for example under linux, you can set the priorities of the kernel so that an incoming midi byte will bypass everything (even harddrive reads) and wake your applicaiton up.

The other problem is with USB, which has a weird architecture where the PC has to poll the slaves (the MIDI device) to ask for data, and then the data gets sent real fast. What that will do is that when you receive MIDI, and a lot is going on on theMIDI bus, the MIDI device may fill data into a buffer and send it all at once over USB, which will result in "blocky" midi, cause a whole block of a few notes gets sent at once. MIDI uses normal BULK transfer on USB, and not the better suited (in my eyes) ISOCHRONOUS transfer which would make the line "tight" (ISOCHRONOUS is used for audio for example).

Also, once the USB packet is received by the PC, it gets reparsed, separated into midi messages, and then gets sent down the midi stack (which is pretty cmplex in both windows and macosx). That's way more complexity and possibilities for jitter than the old ATARI ST code which would basically a function as soon as a byte was received.

There is no real prevision to make anything tight over USB midi, like there wasn't in original midi. Also, before sending a message from the MIDI device to the PC, the midi device has to "parse it" because it gets encoded differently on the USB side. So it has to wait for the whole message to be received (that will not change much on the side of "tightness", but will make for more latency).

Also, midi is 31250 bps, which amounts to 3125 bytes per second. A midi note is 3 bytes, so roughly (if there is no running status), you get about 1 ms per note. send a chord (4 notes) and a couple cc for pedal or pressure. Woops, 15 ms for a single chord between first NOTE ON and last NOTE ON.

my poly800's step sequencer is outrageously sloppy, if you turn the speed up high enough the led's dim with each note and the timing gets a drunken swagger. i like it better that way, it adds to its lofi charm

but yeah, id like to see someone do a real experiment with this. ive never checked the audiofiles close up, but when i use my hardware synths controlled from the computer it always sounds a little less than tight. the fact taht i'm daisy chaining eight synths off of one usb/midi connection can't be helping things

i really dont have much experience with computers for music but ive never ever seen anyones computer be able to play continuous 384th notes x14 sounds at once without slowing down or missing notes out or crashing like my hr16 can - ive set it to output midi notes like this aswell- it may well have varied a little but still thats a fuck of a lot of notes per second- certainly more than the input buffers on my other equipment can handle

i'm glad dawseen wrote all of that because, I didn't want to.
; )

1 | 2 | 3

Register / login
You must be a member to reply or post. signup or login