beats + tempo + soundbites = what am i doing wrong?

[otalgia-2000]

hello everyone

i'd like to be able to tighten up MIDI note and audio timing on several tracks with respect to a specific track in a sequence (rather than locking to the note grid), as well as just generally getting more familiar with the world of quantize, particularly groove-quantize.

the incredibly helpful series of youtube-torials on on beat and tempo detection really helped me get started, but working further with these features has yielded some unexpected results so i'm asking yall.

the beat detection and tempo analysis concepts are well laid out in the tutorial videos. however, those examples are based on several tracks of contiguous audio.

but more often my tracks are made up from a patchwork quilt of several soundbites -- some of which might be edge-edited or have overlapping-layer edges.

i put together two tracks of material in my usual stitched-together fashion and found that the click did not hang in faithfully with the ebb and flow of the timing -- it fell in and out of following the music.

there are a few steps involved in correctly generating a tempo map and i don't doubt that i may have messed up somehow.

but i even went so far as to merge the soundbites in each track (tho i felt merging really ought not be necessary or might at times be not-necessarily-desirable) and re-analyzed that, and was pretty surprised when even that didn't work.

any thoughts on beat and tempo detection and further quantization when you're trying to frankenstein a tune together would be most appreciated!

oh, BTW: my two tracks were as follows:

1. a very clean electric guitar DI and
2. the a further-down-stream DI of the same guitar, but post-stompbox chain.

i used the clean DI track as the source of copy-beats for the other track. the two tracks both belonged to an edit group, and were composited from a handful of soundbites that were created by copy-and-pasting segments from an original longer selection.

thanks, as always!
live from the polar vortex,
O-2k

NOTE: re-posted from original mistakenly posted on the DP windows board!

[Shooshie]

I'm kind of shooting in the dark, here, because I don't know exactly what steps you took, but after moving soundbites, did you tell DP to adjust the soundbite to the tempos in the Conductor Track?

As it works out, I usually start a project either in MIDI or Audio, rarely both. If I start it in MIDI, I'll have barlines and beat information, along with constant tempos in the Conductor Track. Later audio is timed to the MIDI, so there's not much call for me to work with tempos in the Audio tracks.

However, if I start in Audio, I rarely line up barlines and beats, but simply work by ear in the audio tracks. Consequently, my beat-detection skills are not up to par with my other skills in DP. So, I can't tell you if DP automatically adjust beats in a soundbite when you move it, but I THINK you have to tell it to do so.

That feels incredibly un-helpful, so I hope I didn't make matters worse, but if you haven't tried that, it may be at least part of the answer.

Shooshie

[otalgia-2000]

I'm kind of shooting in the dark, here, because I don't know exactly what steps you took, but after moving soundbites, did you tell DP to adjust the soundbite to the tempos in the Conductor Track?

thanks so much for helping out.

in answer to your question yes, i did that.

the steps i took (FWTW) were:

1. analyze beats in the clean DI
2. analyze tempo in the clean DI
3. copy beats from the clean DI to the FX DI
4. select all
5. adjust suquence to soundbite tempo.

the tutorial vids mention the synch point created for soundbites once you analyze their tempos, and suggests quantizing all the tracks in question so as to get everything lined up to measure 1. and that does work in the video.

however what i saw with my multiple soundbites trained together was that said "quantize for line-up" moved each of them a little differently and they lost their precise places in my patched-together arrangement as their synch points looked for the nearest measure. that's what led me to try merging the SBs.

working on it more this evening, i re-recorded the idea as one contiguous track in order to try to reduce the variables. having still less luck than i'd like, i then realized that i may have built my self-tutorial speedbump: my piece has an anticipated feel, so lining up the "one" with respect to the real downbeat and measure 1 isn't quite the cinch it ought to be, and that's been vexing me a little.

As it works out, I usually start a project either in MIDI or Audio, rarely both. If I start it in MIDI, I'll have barlines and beat information, along with constant tempos in the Conductor Track. Later audio is timed to the MIDI, so there's not much call for me to work with tempos in the Audio tracks.

However, if I start in Audio, I rarely line up barlines and beats, but simply work by ear in the audio tracks. Consequently, my beat-detection skills are not up to par with my other skills in DP. So, I can't tell you if DP automatically adjust beats in a soundbite when you move it, but I THINK you have to tell it to do so.

i'm not sure either, so to be on the safe side, i've been continually re-analyzing for beats and tempo, and re-adjust-seq-to-SB-tempo'ing.

and you're right, i am working in a hybrid audio-with-MIDI environment, so techniques may vary somewhat when mixing domains this way. but i feel like somewhere, somehow, this really does kind of work, and i'm trying to spare myself at least some of the happy hours of microsurgery needed as i correct the timing of, for example, pad-triggered MIDI drums to a real live guitar played less than adroitly.

in the past i've just gone on a note-to-note lineup and man, there has got to be a better way to get at least most of the way there (...who was it that said "she won't drive you crazy but she'll drop you off close enough to walk"?)

That feels incredibly un-helpful, so I hope I didn't make matters worse, but if you haven't tried that, it may be at least part of the answer.Shooshie

not at all -- most helpful, and very much appreciated.

o-2k

[Shooshie]

Here's a Magic Dave post on beat detection that I copied to the DP Tips Sheet Thread, page 7. It may tell you nothing new, but I think I'd give it a chance on the possibility that it has some nugget that makes the difference.

Shooshie

[otalgia-2000]

thank you very much, shooshie -- this was very helpful, as can so often happen when you get the chance to see information expressed in a different way. sorry it took me an extra couple of days to read it and get back to you, but i just had to jump in and try it first.

my own tutorial example seemed to benefit best from using "adjust beats" which, to be sure, is laborious. but of course, once you have the sequence tempo mapped, then any subsequently-recorded track will benefit from the time you put into the adjusting, which beats by manually adjusting any new track that needs it by about a case of light-years.

and yeah, i might could have picked something a little more generic to try learning this stuff on, in which case the more automatic functionalities are there to call upon.

thanks again,
o-2k

[otalgia-2000]

further to the topic:

after continued research i'm starting to form the impression that it helps to develop a feel for what material will do fine with simple auto detection, as opposed to the kind of stuff that's better served by just going ahead and recording the beats you want.

after some effort i realized that the tutorial example i've been using for my experiments turned out to be Type B -- it just took me a little time realize it. but it was time well toiled over: a simpler example might not have yielded this critical distinction, and that was instructive.

however i have now at last created my first (*really!*) crude tempo map. now: will i use these powers for good?

...or for evil?

only time will tell (but i can sure tell you, my time aint great. thanks, MOTU -- there's nothing quite like the sequence editor to show you how far off you are...)

o-2k