While it's not officially supported,
it is apparently possible to alter the installer configuration to have that feature work on on-RTX graphics cards, so if you have a relatively recent NVIDIA GTX series card you might be able to try it for yourself.
Note that as this technology actively manipulates the sound coming from your input device, if you do get it to work, the speech recognition engine may become (much) less accurate.
Speech recognition training is partly intended to adapt recognition to your microphone and environment, but if the NVIDIA software is continuously changing how your microphone sounds, that could diminish the effectiveness of that mechanism.
I would recommend
creating a new voice recognition profile and training it (three times at least) to check whether that does improve accuracy.
If you find the regular speech recognition engine has issues with recognition while this software is active, even after re-training, you could try the
alternate speech recognition engine, which comes with a number of caveats (mainly a lack of dictation support, including any feature that relies of freely recognizing non-predefined speech), but cannot/does not need to be trained, and may be more suited to operating under conditions like that.