Your'e probably right that its the speakers.
Let me be clear, though, I really liked the NAD M33 and think it is an excellent sound, but the Elicit-R with the SFs just sounds better than the M33 with the SFs. I am talking about the overall combination, not amp vs amp, which would be hard to do as I am not sure you can bypass the M33's DAC (?)
To kind of emphasise your point - I think I should have upgraded my speakers and not the amp. That's what I am hearing.
So Gumby > Elicit > KEF vs NAD M33>KEF I made the decision to change. Later added the SF's to the NAD - better still. By chance because of the whine/buzzing noise in the M33 that meant it's gone to the service centre, I discovered that the Gumby>Elicit >SF sounds the best of all combo's. (of course, without side by side / blind tests.. though I do listen for hrs every night purely focused on the music)
On DIRAC - I tried it and I found that using it actually lowered the dB by about 7-8. Not a major problem, but I also found I could get better results via Roons DSP prior to sending the data to the M33. I am sure DIRAC experts could do better than me, though.
Getting Dirac right takes a few attempts and has a bit of a learning curve. What should be a simple measurement exercise can easily turn into a mess with a few seemingly insignificant mistakes. A few measurements taken too close to a wall, or the main listening position being slightly off can have a big impact.
However, something not often talked about is that the positioning of the speakers is a very important starting point for a successful Dirac measurement. If the speakers are placed so that there are large nulls or large room modes, Dirac will do its best to correct, but you may lose gain or even get too much.
There is software called
REW that I think should be required learning for all audio enthusiasts. In fact, for every dollar wasted on useless high end cables, I wish the author of REW received even 10% of that. He would become deservedly rich, and the end user would actually get something of value.
When you get the M33 back, I would suggest the following process:
1. Use REW to measure the speaker placement and try different positions to smoothen the frequency response as much as possible. You can use foam plugs supplied by the speaker manufacturer to test different bass responses. If the speaker is near a corner or back wall, try to move it away from the wall to smoothen the bass response.
2. Run Dirac measurement and apply a Harman +6db curve or the NAD Room Curve from
https://nadelectronics.com/dirac-live/
3. Use REW to measure the before and after with and without Dirac Live engaged.
If all goes well, you should have an after frequency curve that is relatively smooth (it will never look as smooth as the after curve in Dirac because that's based on an average of several measurements vs. one measurement in REW from the main listening position. You can take an average of measurements in REW from different spots as well using a moving mic method, but suggest doing just 3 measurements to keep things simple around the main listening position.
After you've got a smooth frequency response, take some time to adapt to the "new" sound. It may sound flat and a little lifeless at first, but once you become accustomed to the nuances in the bass notes, you'll never want to go back to the intense room modes and nulls. It took me about 2-3 weeks to adapt to the new sound. I will admit however that I still keep a +10db Harman Curve in one of the filter slots for when people come over and need to feel the excess bass to enjoy music
Good luck with your M33. Hope NAD sorts out the whine issue. While my M33 isn't perfect, it's the best combination of features, power and looks on the market that I was able to find.