Tempo change marker location quantization

If I am not missing some settings or options, it seems that Ardour is quantizing the location of tempo markers imported from a midi file to quarter notesl. This happens independent of the time signature.

It also seems that when creating tempo markers manually, the tempo marker locations are also quantized to quarter notes, but then can be moved to beats if the time signature provides for beats shorter than a quarter note.

Quantizing the tempo changes to beat locations is sensible. And given how the midi standard is focused on quarter notes, assuming that the tempo changes will happen at quarter notes is understandable.

However, the midi standard does not limit ( to my admittedly limited knowledge ) the location of the tempo markers to quarter notes and it is very much possible ( although likely not good practice ) to create them at any instance in the time line with a resolution of ticks.

But there is ( to my understanding ) a mostly unused midi parameter, “notated 32nd notes per beat”, commonly set at 8. Perhaps this parameter can be used to set the quantization of location of imported tempo changes. Or, perhaps the actual time signature determined beat locations can be an option.

I understand this does not affect most people. But there are midi files out there ( mostly classical music ) that use this finer tempo change location quantization, meaning there likely is a software that allows this finer quantization.

I also guess this would likely be a non-trivial change in the Ardour code. Just wanted to mention and get some feedback.

Also want to mention that ( at least in my experience ) with the recent fixes at commit b76c3b1 ( Nov 17 2023 ) Ardour now handles midi files more elaborately and consistently than any other open source ( and likely any other commercial code ) out there. So, thanks a lot…

Unfortunately, Ardour does limit tempo changes to beats (quarter notes if the preceding tempo is in 4). Changing this not possible.

We consider any other possibility to be musically non-sensical. You cannot change tempo (discretely) between beats.

The problem comes principally because SMF has no method of indicating ritardando and accelerando explicitly. You can only put in a series of discrete tempo changes, and the SMF creation tool may not quantize these to be on-beat. Ardour does, and the difference should be minimal. This is also one of the reasons we do not “export” our tempo map into SMF’s created by Ardour (for now): there’s no coherent way to represent the way we do ramps (which models human performance).

For non-ramped discrete changes, I have yet to see an argument that justifies them being off-beat.

Thanks for the clarification Paul. I was suspecting that this all was due some software’s attempt at exporting ramped tempo to SMF.

As you point out, the difference is actually minimal. And it is also clear that changing tempo off-beat does not make sense.

Thanks for taking the time to explain…

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.