I had an “Oh, WOW!” speech experience with my Sphere I90s today. I went to Walmart to get a COVID booster and a flu shot (I also found if you’ve had an RSV shot, you don’t need another one).
The pharmacist who showed up to shoot me up had a pretty deep voice, and the edges of it blended in with the background din of Walmart, making his speech a bit muddled for me. Even though I’d set the noise trigger level for DeepSonic processing to kick in at only moderate noise, it didn’t kick in, and nothing happened (maybe I didn’t let him get enough speech in). I switched to my manual Spheric Speech program. Bingo! Big improvement. I could understand him, but his voice didn’t stick out amazingly from what was left of the background din. So I tapped Adjust Program down at the bottom of the screen and tapped the SPEECH option. His voice then clearly stood out from the background remnants, almost like Speech Enhancer had kicked in. I saved the Spheric Speech with the SPEECH option as a new program. And I felt, “Oh boy! These I90s are worth the price I paid!” 
My experience was hardly a controlled experiment; maybe some variation in the background noise led to the pharmacist’s enhanced speech. I had another experience in the same visit, standing by the entrance to the Walmart store. My HAs were in the Automatic program, probably in Calm or Calm in Noise. A bunch of Hispanic guys walked in, talking loudly in Spanish. Bingo! Sphere noise reduction automatically kicked in (the first time I’ve ever been in the right environment for that), and I could clearly hear the Spanish above the din. It made me wonder what SNR ranges of speech to noise can automatically trigger Sphere noise reduction. There must be a Goldilocks zone. With an SNR that is too low, speech probably can’t be detected by the HAs in the noise; with a speech volume that is too high, you have a good SNR and don’t need Spheric processing.