So I have a new song I’m working on. It’s been one of my more ambitious ideas as I wanted to implement tempo glides into the song, but… there’s been multiple issues with trying to implement that.
The first thing is at first there seems to be an obvious solution; just put SynthV into your DAW and it’ll sync up! But, there’s an issue with that. I only have SynthV Lite, and Lite only functions as a separate program, you can’t just boot it up into your DAW. And even if you COULD boot it up into a DAW, I wouldn’t be able to implement all the tracks because my song has 5 vocal tracks in total and you can only have 3 vocal tracks active at a time in SynthV lite.
I thought I would be able to work around this issue by composing a MIDI file in a different program, export that MIDI, add lyrics to the MIDI in a different vocal synth editor such as OpenUTAU, convert that export into a SynthV project file, and then send that file to someone else who does have SynthV pro to handle the rendering. But there’s ANOTHER problem with that. In the case of the DAW I’ve been using (LMMS), the DAW isn’t able to incorporate tempo changes into a MIDI export. Now, I could just add the tempo changes manually in a vocal synth editor, but there’s another problem; you can’t really create tempo glides in vocal synth editors, which is a problem because my song DOES have tempo glides. If you wanted your vocals to sync up with the song, you’d have to be incredibly precise about the tempo changes, or else everything would eventually go off-tempo.
If anybody has any help and/or suggestions regarding these issues, that would be much appreciated.
In what range (bpm) are the tempo changes?
Unfortunately, I don’t know the DAW LMMS, as I mainly work with Cubase.
So here’s a suggestion as to how I would proceed if the tempo changes are within a manageable range (only slight ritardandi or accelerandi).
In this case, I would record the voices in Synth V Lite at a fixed tempo, export them as audio files and then import them into the DAW. The DAW must also first be set or switched to the same fixed tempo.
If the imported audio files are then switched to an “elastic” mode in the DAW (in Cubase this is called musical mode), they will follow every tempo change.
Of course there will be some loss of quality or artifacts on the audio tracks if the playback tempo is not identical to the original tempo from Synth V, but if the elastic algorithm is of a sufficient quality, these effects are not very significant - as I said, the whole thing depends very much on the strength of the tempo glides.
I see another possibility that does not require a fixed tempo if you work with step-like tempo changes in Synth V, even though they are programmed as ramps (glides) in the DAW.
As you have already described, the audio files exported by Synth V are then no longer played back exactly synchronously in the DAW. However, this can be remedied by cutting up the audio events and moving the individual audio snippets (called “events” in Cubase) to the correct positions.
A method I have used is to create a backing guide track in your DAW or what ever with the required tempo changes, render it as a .wav and import into SynthV, use that to set your tempo changes as you proceed with adding a vocal track. I changed tempo on every beat at one point to keep vocal synchronised to backing but you might get away with correcting once per bar.
Render the tracks as seperate files so you can dump the guide track and import the vocal back to the DAW for finishing.
Maybe not ideal but I have one song that varies from 60 to 93bpm with this technique.
The tempo changes in the song were as follows:
-Initial tempo is 102 BPM (with 6/4 time signature)
-tempo changes at… bar 20 to 153 BPM? (I’m not sure what the exact term is)
-tempo changes at bar 28 to 102 BPM again
-from bar 42 to bar 44, tempo glides from 102 BPM to 153 BPM
-from bar 48 to bar 52, tempo glides from 153 BPM to 204 BPM
-from bar 56 to bar 60, tempo glides from 204 BPM to 77 BPM
-from bar 68 to bar 72, tempo glides from 77 BPM to 102 BPM.
I did make a screenshot of my work on the vocal midi for my song to hope show what was supposed to be going on:
OK, so in principle its the same as the song I described, more tempo changes but the same scenario.
As you only have SynthV lite, any scripting options are out but they wouldn’t totally solve this problem anyway.
LMMS I have never used so may be misunderstand your use case here but are there instrument tracks in this song or is it just a cappella? If you can produce an instrumental track as a guide to import into SynthV then do as I describe above, produce a couple of the vocal tracks and render them as wavs, delete the vocals BUT KEEP THE INSTRUMENTAL so the remaining one still has all the tempo settings and add a further vocal track or two, repeat until all five are done.
No, the song isn’t just a cappella. It’s just that I decided to make the notes for the vocal tracks separately so that they would be easier to export. However, LMMS does have issues with its MIDI export where it just doesn’t export tempo changes to the MIDI file, even though I’ve seen it’s very much possible to include tempo changes and even tempo glides into MIDI files.
Import instrumental tracks as .wav files, add all vocal tracks in SynthV from scratch or accept that imported vocal MIDI data will have timing errors that have to be fixed note by note, at least the pitch data is correct - or try exporting seperate MIDI files for each fixed-tempo part of the song and stitch them together in SynthV?
Your explanations doesn’t sound all that helpful. It all seems to just loop back to “oh just do it manually” like do you have any idea how next to impossible that is for a human to do? In a tempo glide, there could be multiple tempo changes occurring PER NOTE, and the exact time those tempo changes occur can be very difficult to determine.
I swear there has to be some program out there where you CAN compose a midi featuring tempo glides in it because the official MIDI for “CHRONIC WASTING DISEASE” by R.I.P DOES have a tempo glide in it.
Hey Deadly-Espresso that’s a little uncalled for: you indicated the DAW you used has some limits that mean Mechie’s suggestions are the only ones that seem possible to Mechie. Mechie hasn’t said get another DAW or said anything other than try and make helpful suggestions.
This resolution of tempo change is NOT NECESSARY, who can tell that the first half a note is faster than the second half?
A beat may be two or more notes, all at the same “tempo” as seen by SynthV but that can easily be worked around when tuning, tweaking start and duration of notes.
You have some pretty powerful software at that price point there, LMMS is free and SynthV lite is free, even with paid-for software I am doing it manually.
What do you mean “this resolution of tempo change is NOT NECESSARY”? If you have the wrong tempo for even less than a note it’s going to desync at some point. The tempo changes need to match up perfectly with the song or it IS going to desync at some point.
Another possibility that takes into account the fact that Synth V only allows tempo changes per quarter note and the tempo values must be integers (without decimal places).
How about programming step-like tempo changes instead of glides in the DAW?
You could then transfer these 1:1 to Synth V (you have to enter them manually).
The difference between step-like and continuous tempo changes should be barely audible if the changes are sufficiently close together, i.e. in the quarter tone grid.
And finally, another tip: You can also calculate the necessary step-like tempo changes for Synth V to match them almost perfectly with continuous tempo changes in the DAW.
@Deadly-Espresso I mean EXACTLY what I say. If you feel it is important you can resynch at the start of each note but it works just fine to work at bar intervals.
I think you are imagining some sort of cumulative error, slightly too slow on one note over ten bars will drift your timing but that is NOT what I am saying. @Pat is describing this and appears to be getting away with it …
EDIT: @Pat Just in passing (not tested it myself) fine_tune_bpm.lua
Set tempo to 0.1BPM resolution?
How did you calculate this? Also, the time signature is wrong because the song’s time signature was 6/8, not 2/8.
Edit: I think only changing the tempo in whole increments (not in decimal) COULD work because from what I’m observed, LMMS can’t handle decimal placements in the tempo values. However, I’m still unsure as to if the numbers you calculated will still succeed in making a vocal track that syncs up with the song.
Okay another update, I did try out the values suggested in Pat’s spreadsheet and… they’re okay, I guess? They mostly sync up, I think? They’re better than what I expected, that’s for sure.
Thanks for the link! I’ll take a look at the script.
In the svp files created by Synth V, the tempo information is actually saved with one decimal place. If you manually change the tempo values in this file, Synth V will also adopt them correctly!