The Xceed uses the OpenSound Navigator, which is the noise reduction technology for the OPN and the OPN S. It does noise reduction differently than the More/Real/Intent. The OpenSound Navigator cleans up the speech diffused with noise by creating a noise model of sounds from the sides and rear and subtract this noise model from the speech in front that is diffused with the noise (that is represented in the noise model). This helps give a clearer speech that is minimally diffused with noise.
The More/Real/Intent uses the AI DNN (deep neural network) to break down sound components from the sound scene and recreates it. And in the process of recreating it, it can rebalance the sound components to give speech more volume and non-speech sounds less volume as a way of achieving noise reduction and still be consistent with the Oticon open paradigm.
Initial reviews on the Intent seems to be positive on speech in noise compared to the Real. I think this is 2-fold: 1) Oticon is able to eke out an additional 2 dB of SNR for speech, which hopefully will make speech stand out even more, and 2) the DNN 2.0 is trained with more diverse sound samples, which hopefully improves the accuracy of the sounds characteristics (sounds seem more real).
Having explained the 2 approaches (of the OpenSound Navigator or OSN in the Xceed and OPN/S vs More/Real/Intent), it’s more obvious that the OSN “cleans up” the speech from diffused noise, while the DNN rebuilds the speech components so that the emulated speech is “originally” clean already.
Now back to your question regarding max NR, whether it’s always good to set it to the maximum available value or not. After all, you pay a premium for the top tier aid, why not take advantage of the best NR you can get? Evidently if it’s always good, then Oticon would not have given you a choice of setting what the max NR is, but instead would just always apply the max regardless. But obviously there is a trade-off in scrubbing the diffused noise from the speech, because if you scrub too much of the diffused noise, you’ll scrub into the speech components that are masked by this diffused noise as well. Kind of like scrubbing bugs off a car finish. You get too abrasive, you might get through the clear coat and maybe even down to the paint.
So I’m guessing this is the effect that you see with the Xceed. If you have it set at -9 dB NR max, it may scrub too much diffused noise off that you lose some integrity of the speech component that gets scrubbed along with the noise. If you set it at -7 dB NR max, you don’t scrub off the noise as much, but at least the integrity of the speech might not be compromised as much. There is a trade-off here depending on your tolerance of the noise level, to be balanced with preserving the speech integrity. And that trade-off is left for your HCP and you to decide.
It’s also worth clarifying that the max NR setting Oticon gives us is not the “always on” flat out NR that is applied for any situation. Oticon has a way to determine the complexity of the environment in real time, and applies the NR only proportionally to the actual level of complexity at the moment. For example, if you set your max NR to be -5 dB and you have 6 scenarios of increasing complexity, S1 being the simplest to S6 being the most complex. Then if -5 dB max NR is what you set, then S1 is given 0 dB, S2 given -1 dB, S3 given -2 dB, so on until S6 is given -5 dB. But if you had set your max NR to -9 dB, then S1 would be given -1.5 dB, S2 given -3 dB, S3 give -4.5 dB, S4 given -6 dB, S5 given -7.5 dB, and S6 given -9 dB.
Of course this is just an example based on my personal guess, and perhaps it’s not that linear like I presume it to be. But at least it gives an idea of scale, to point out that Oticon applies only the appropriate NR to the appropriate complexity level, and Oticon gives you a crude option to put a ceiling on what the max amount of NR you want. This goes back to how well you can tolerate the noise. If you can tolerate it better, you set a lower max NR limit so Oticon won’t scratch too deep as to possibly mar the speech surface (so to speak). But if you cannot tolerate the noise well, then you have choose to have Oticon do deeper scrubbing of the noise, but at the expense of marring the speech surface too much and lose some integrity of the speech sound.