Page 1 of 1

again quantize ,more clearly (i hope)

Posted: Mon Mar 17, 2003 11:33 am
by miky
yep, maybe i was a little unclear but played last night again with the demo and :

quantize in ableton is in fact a latency induced in playing the clip :1 bar,1/2, 1/8 etc. so if i want to quantize 1/16 and i play the clips i will hear a latency of 1/16 etc.
to avoid this :
i just need to play the clips without quantize enabled (i choose to disable the quantise) to avoid the ableton quantize latency, i play them ,everything sounds ok but after that i need to quantize the clips like on a regular sequencer (steinberg,logic etc.)
i hope this is all clear now,
and it's really nice to see you answering monolake ,this reminds me that i made the right decision concerning "going LIve route"

Quanta quantum quantos

Posted: Mon Mar 17, 2003 1:58 pm
by monolake
hi miky,

now i think i do understand your problem.

(un) fortunately Live combines two different worlds :

a) the abililty to play clips in sync in realtime ( e.g on stage or when plying with other people )

b) the abillity to record these actions in a timeline.

in b) quantisation in a MIDI sequencer usually assumes that a note playing "around" a specific time is meant to be on that time :


note a and b are quantised to 1 while c and d belong to 2.

in situation a) clip a will start at postion 1. but clip b cannot start at 1 because Live is not knowing that it will come. the only usefull thing Live could do is to start the clip at position 2. ( which would not happen in the above example since the user did decide to choose clip c later... )

( actually the behaviour is like changing patterns at a drum machine then playing notes )

since the arranger is a protocol of what you hear in a realtime situation
it has to behave accordingly and this is indeed different from what a MIDI sequencer would do.

regards, robert / ableton


Posted: Tue Mar 18, 2003 6:46 pm
by ZeroAltitude
This is very interesting to me. Is there a technical reason Live could not be told to behave like this, in real time:

1) when 'b' is triggered, and 'real time quantisation' is set, do the following:

a) note the time 'b' was triggered
b) calculate the most likely point at which 'b' should have started, given the quantisation setting (e.g. if quantisation was 1/4, find the nearest quarter-note, even if it were in the past)
c) calculate the (possibly negative) offset
d) if the offset is positive, wait til that time to start 'b'
e) if the offset is negative, play 'b' immediately, starting from time (START-B + offset)

I am intrigued by this question. Your answer would be most appreciated!



Posted: Wed Mar 19, 2003 11:34 am
by monolake
.. the behaviour you describe would lead to very unpredictable results. imagine that you trigger a rhythical loop. if it comes a little bit too late it would be complety out of sync. i assume this would cause lots of irritations.

as i said in my last post, the metaphor of live is closer to a pattern based sequencer then to a midi note recorder. none of these concepts are right or wrong at the frist place. it´s just two different things and we believe that for playing loops live the pattern model is the better approach.
this does not mean that iwe would implement a midi seqeuncer the same way. the midi sequencer obviously works best when capturing events around a specific beat position.

regards, robert

// personal statement : difficult to talks about music and sofware these days while the amerikan dictator is going to attack the iraqian dictator ( and thousands of innocent people ). the iraq could become so "powerfull" thru millitary support by the USA not so long ago ( if you can say a rotten poor country like iraq is powerfull ) . george bush is ignoring international laws. george bush is _attacking_ another country. almost all criteria bush uses to justify this attack are also true for the USA. the USA has wappons of mass destruction, the USA is ignoring international laws. the USA is an agreesive, unpredictable, dangerous country, the USA is not telling the truth and so on and so on. who is more dangerous for the world ? a cracy dictator of a small underdeveopped country or Geroge Bush ??? //

Posted: Wed Mar 19, 2003 11:17 pm
by ZeroAltitude
I am only following this out of interest, not because I think the Live paradigm for non-MIDI beat-syncing is wrong or anything. I'm just curious.

Why do you say the behavior would be unpredictable? Is it because the real-time calculation of the offset into the triggered loop would either be error prone, or too slow to make the loop predictably sync'ed to the other outputs?

If things were infinitely fast, then I would say, you could (in principle) know *exactly* how many microseconds into your triggered item 'b' you are, given that 'b' should have started y microseconds ago.

But this is why I was most interested: are there practical programming or hardware speed (today) limitations that prevent this? Or is it just, that you think the feature if added would make the UI too confusing/cluttered/unintuitive with the additional controls/options?