Hi there! You're graduating? That's awesome.

I'm a regular on CodeReview SE but I just posted a question on your site. Just figured I'd stop by and say Hi!

Hello! Yes, I just heard the news -- very excited!

We haven't graduated yet but been working hard at it

Just glancing at your question, that's something a high quality sequencer or DAW software should be able to do

what's the software you're using?

Propellerhead Reason 7

Hmm, I'd expect Reason to have that capability, but I haven't used it myself since version 2 or 3

Problem is, when you import a MIDI, it imports the assigned tempo with it. In this case, the notes are misaligned to the tempo...

So if you try to quantize... well... it goes by the assigned tempo, which basically sounds like garbage

right

are you certain it's consistently off from the assigned tempo?

Yeah. It doesn't seem that the performance is even consistent, but as far as I can tell, the music drifts off (as being slightly slower) as opposed to the click track

because if you wanted to zero in on exactly the BPM value, you could basically count the number of beats from a start and end point and divide by time

*divide time by that number of beats

on an unrelated note, your question gave me the idea that we should probably have a reason tag -- if only because there's about 1,000 question results for that word, but only a small percentage about the actual software

Hm, another idea for your MIDI issue -- might be a bit involved, but if there's no technically easier way, I could see using the MIDI file as an input source for some kind of reasonably intelligent notation software like Sibelius -- in that it might detect and compensate for "human error" that's really just what's in your MIDI file

Problem I'd be running into, if I did that, is that the MIDI notes still would be misaligned; though I might be able to figure out how to scale the tempo appropriately

(right side)

I would expect that if the notes are just slightly off but the overall tempo is consistent, then setting the right tempo would allow you to quantize with a high degree of accuracy

that is, if tempo is consistent and interpreted correctly, then the note "margin of error" stays within a confined range throughout the track instead of getting more and more off base

It wouldn't unless I scaled the notes to fit it; but yes, I'm guessing that would help a great deal

One more question, if I may...

For the sake of accuracy... would I count beats all the way up to the very beginning of the last beat (say, the last "thump" at the end) and time in the same fashion, to get most accurate result?

you wouldn't count up to the last beat, but you wouldn't count it

@Phrancis The real question here is how you managed to get the MIDI track to play at a tempo that is not the tempo of the track!

if you like, think of what would happen if you had a piece that was two beats per minute

I am genuinely surprised you don't have a reason tag, being one of the most popular VRS out there

a beat is just a point in time, so if we look at three beats, at 0:00, 0:30, and 1:00, you only measure complete beats in the time range

so between 0 and 1 minute, you had two complete beats

@JCPedroza I didn't, the person who made it did. It's basically "free hand" recording, ever-so-loosely following an approximate tempo

(as far as I can tell, I hope I'm wrong)

err, I think my wording was a little off there -- a beat starts at a point in time and ends at the next beat

@Phrancis Your question mentions a non-integer tempo. I'm assuming the track was not recorded using a click/metronome? The reason the tracks are not aligning might be because the actual BPM of the track is not constant, and you were given the average BPM, not the constant one. The MIDI track, on the other hand, once quantized will have a constant BPM.

@JCPedroza Big question mark for me as well. I'm fixing another "producer"'s crappy work.

@Phrancis Just check the percussions, the rhythm. Use a grid that uses measures, beats, etc. Do the elements align consistently with the grid?

Chances are that they do not. Happens all the time with songs recorded without a metronome or click track

I'll tinker around with it. If I figure it out, should I selfie answer or delete question?

The imperfections in the tempo make the bpm not constant

Don't delete, it's a common and useful problem

Some people just assume that everything has constant tempo, but many bands and projects in general don't use a click track or metronome, which causes the tempo to vary through the song

i mean it's a common problem and the question is useful

Gotcha.

If you still can roll back to the not quantized MIDI, do it

Yeah, not an option. I'm just handed this crap MIDI file and being told, "fix it."

If the tempo is not constant, which seems to be the case, you'll have to fix any mistakes manually (which is not a big deal in the MIDI realm anyway)

oh haha

I agree -- the solution to your problem has a lot to do with why the MIDI file is the way it is. Unfortunately, if someone just recorded their own manual playing without a click track and didn't bother to clean it up, you're mostly screwed

I think they did have a click, but were not very good at playing along with it...

Well, assuming the song doesn't have a constant tempo (which it's almost always the case with non-integer tempos like 120.4 or 110.98) and that MIDI file was quantized (it has a constant tempo), you'll have to align everything manually

Hopefully I'll be able to coerce the MIDI into fitting (mostly) into a tempo. Otherwise, I'm in for some fun editing.

well not everything, just the MIDI track

Even if they are not good an playing along with click, things should align if they used it

You might actually have more success breaking the track up into much smaller chunks that CAN be quantized successfully at their own tempo -- then stitch the chunks together later and decide on a single tempo for the whole thing

It's 21 tracks - I hope there is an easier way to fix than by hand

You just need to fix the midi track

Underlying...?

Like tempo-map or something?

click and drag

click and drag the midi notes to get them where they should be

(haven't dealt exceedingly with MIDI files, I usually write the music)

I think it's 21 MIDI instruments in a sequencer... are they ALL inconsistent in different ways?

^^ They seem to mostly line up rhythmically (drums are a mess though)

(note-wise / unrelated)

Wait, are all tracks MIDI tracks?

Sequenced MIDI tracks or recorded MIDI tracks?

There are no recorded audio tracks?

There's a "reference" audio but that one is unimportant, could easily do without. Everything else is recorded MIDI (as far as I can tell)

ouch

Have you tried quantizing everything to 16ths?

well, at least the chunking option could still work -- you'd just do all 21 tracks at once

@JCPedroza I thought about it, but look at the 2nd screenshot in my question

Notes that are behind would just stay there, 1/2 beat off tempo

While they should be 1/2 ahead right onto the first beat

quantizing just glues everything to the closest grid it doesn't know what the performance wanted to achieve

That's the core issue; (maybe I should edit question): The MIDI is way off.

Here's another approach that might apply in this context -- when I transcribe music like a jazz solo or something, I use a piece of software called Transcribe that lets me place beat and measure markers on the timeline. I think that should really be your first task: figure out just how constant or not the tempo is throughout the entirety of the piece

I was hoping it would be able to disassociate the MIDI notes from the MIDI tempo grid, and set them up differently

Why not quantize and then move the track to compensate? if the midi is 1/2 beat off, why not move the track by 1/2 beat and align it?

You can move the MIDI notes wherever you wan't you don't have to follow the grid, the grid is just a guide

@JCPedroza it's just not consistently off, it just drags further away from each other as the song goes

Then the problem is not that it's 1/2 beat off, the problem is that the tempos don't match

But that doesn't make sense

@NReilingh would you provide a link? This may help

All tracks have the same tempo

All tracks are at Reason's master tempo

Exactly. If I may sum up my issue in a few words... it feels like the performer thought... "LOLTEMPO"

Have you tried quantizing all the other tracks?

the other 21 tracks?

They're all off about equally (copy & paste)

that's irrelevant

@Phrancis It's this, but you can achieve the same thing with reason -- you just want to place some kind of beat markers on your timeline so you can see for yourself what the tempo(s) are at various places

semi-related: what's the problem with just guessing and checking with various tempos until you zero in on what the music actually is? That shouldn't take too long.

Problem is getting the notes to fit the MIDI tempo map... but I guess I can figure out a workaround, otherwise I'll try to figure out how to slice it, probably with ReCycle

I'll post a selfie if I figure it out

That's also basically the equivalent of using a metronome with a tap tempo function as you listen to playback

Right, just more accurate right?

of course

Cool, you guys have been very helpful!

Open question for whoever sees this: now that graduation is planned, should we re-run the site rename poll? I think we should, as the original question was started just after we entered beta, and the second place answer was created almost three years later. Also, the current users are probably different from who they were three years ago.

If I get positive feedback for this idea here, I'll create the new meta question with the existing options, (including keeping the name the same!) that way I can't vote for my own answer :-)

> By improve, I mean shorten via composition, mapping, folding, or any other best powerful, best practice.

Southern English?

Whoops, wrong room... Sorry...

@NReilingh Do it!

@NReilingh Do it!

@NReilingh Now is definitely the time to do it.

We should also have a "keep the same option" in case people want to keep the name the same.

@NReilingh yes - do it :-)

Working on it now! Will be posting shortly.

And upvoted. You might want to add in a link from the graduation post to make sure folks see it.

Ah, shev made a post a few hours ago about the topic

I closed the old question as a dupe of the new one, so anyone who clicks one of those links should get bounced to the right place.