RapidComposer v5 beta feedback and discussion

[Brian Collins]

A quick experiment with the new Live capability left me very confused. I created a simple 4 measure progression with a chord track, went into live and then tried:

1) Using the allocated Midi keys (C2, C#2 etc)
2) Pressing the Live key letters with the mouse.

Each gave strange loopings which I hope is clear from the video:

For the 1st half of the video I am using Midi keys and for the 2nd half, I am using the mouse.

[musicdevelopments]

I am sorry, Brian, there were a few ugly bugs in beta10 which were fixed since then.
I confirm Live mode works well now, I even created (an unreleased) YouTube video tutorial about it...
Most probably there won't be a beta11 as I plan to release v6.0 in the second half of this week.

Thanks!
Attila

[musicdevelopments]

hey bro nothing news about sdk for project files ?

Thanks for bringing this up! At the moment I don't have plans to add an SDK for project files, it would take quite a bit of work, and so far there hasn't been much interest in it. I do see how it could be useful though, so I'll keep it in mind if more people ask for it down the line.

[sj1]

I plan to release v6.0 in the second half of this week.

Before there is even any beta testing of live mode recording?

[musicdevelopments]

No, I am afraid there is no time for further beta testing. I am updating the user guide and website and do all the crazy administration that is required before a major release...

[sj1]

BTW, is there a list for notification of new official releases? I never get any email for these.

[musicdevelopments]

While I do not have an announcement mailing list anymore (e-mail became unreliable and it is easy to get blocked...) I am going to send e-mails about the new major update to all v5 customers using a reliable e-mail service. Lots of users will be eligible for a free upgrade due to the long grace period, those will receive the new license key and download links as well.

And of course I'll announce "RapidComposer 6" here on KVRAudio and on other websites.

Thanks,
Attila

[sj1]

So, to confirm, no Live Mode recording and no "Loop on Touch" in RC 6.0.0 ?

[musicdevelopments]

Right, I could not finish those. Live Mode recording comes in v6.1...

Thanks,
Attila

[BluGenes]

I suppose it's time to lock this one and start a beta 6 thread. But, before you do that, I still can't get the vst3 to install.

[BluGenes]

so ?

[lulukom]

Hi Attila. I honestly don't know what to do. While the standalone program works fine, I'm having problems with the plugin. The VST3 and AU plugins don't work in MOTU DP 11.35 on Mac OS 10.13.6. I'm attaching a picture that shows DP VST3 and AU as not working.

[heliacal]

No problemo here on Mac OS 13,5,2 with VST3 in Studio One.

[lovemusic15]

I’m deciding whether to upgrade from v5 to v6 and I need clarification on two main things:

Neural‑network / MIDI generation — how different is it from the Idea tool?
From the demos I’ve seen, the v6 song ideas feel random and lack consistent patterns or genre‑specific rhythms. Compared with tools like BandLab (which can generate clear song starters — melodies, basslines, grooves and section changes you can plug straight into basic VSTs), v6 demos don’t show the same level of ready‑to‑use results.
Key questions:
Is v6 still as “random” as the Idea tool, or can it produce reliable, modern song starters (distinct melodies, bass grooves, rhythmic patterns, and sections) that work with ordinary VSTs (not just sequenced VSTs)?
What datasets were used to train the model? If it’s mostly public‑domain/royalty‑free MIDI (classical/old material), that could explain dated outputs.
Is there (or will there be) an option to train or fine‑tune the model on user MIDI datasets (personal MIDI packs, songs you made, purchased patterns)? Allowing user training (even as a paid service or one‑off job) would be a game changer — it would let generations reflect each user’s preferred styles and modern genre patterns.
Bottom line: I want to know the real, practical advantage of the neural network over v5/Idea for producing usable song starters. If v6 only gives random fragments, many producers (including me) may stay on v5.
“Follow Rhythm” → apply to all mono channels
v5’s Follow Rhythm feature is brilliant, but it’s slow to apply when importing full songs because you must change each track individually.
Request: add an option “Apply Follow Rhythm from this Channel → All Mono Channels” (or “Apply to All Channels / Apply to All Mono Channels”). Exclude stereo pads/strings so only mono parts (melody, bass, guitars, lead synths, vocals) adopt the rhythm. This would massively speed up experimentation (e.g., turn a rock track into a country rhythm instantly).
If v6’s neural network can’t yet produce modern genre‑specific rhythms reliably, this Follow Rhythm option becomes essential.
Extra context and why it matters

v5 already covers many core workflows. My upgrade decision hinges purely on whether v6’s AI offers genuinely better, modern, and usable MIDI song starters — not just different random ideas.
If v6 can be trained on user MIDI libraries (paid or local), that would justify upgrading for a lot of producers who want consistent, personalized results.
I’ve attached screenshots comparing BandLab’s generators with your neural outputs to illustrate the differences in structure and rhythm.
Please clarify:

The practical improvements of v6’s neural network versus the Idea tool and v5 for producing usable MIDI song starters.
Whether user dataset training/fine‑tuning is possible or planned.
Whether you’ll add a “Follow Rhythm → Apply to All Mono Channels” option (or similar).
Thanks — clear answers on these points will help a lot of people decide whether to upgrade.

[heliacal]

I don't think ANY "tool" is going to do what you're asking. Why not just learn to write the music yourself and get EXACTLY what you're looking for?