The stuttering literature has been reporting since the 1930's that people who stutter are more fluent when they are alone. But there has been some variation in the fluency levels when alone - some studies found excellent fluency when alone, while others found some level of disfluency when alone, even though the fluency was always better than when in the presence of other people. Professor Eric Jackson (NYU) seems to have reconciled these variations with his 2021 Journal of Fluency Disorders study of "private speech". His conjecture was that in some of the previous studies, although the subjects were \*told\* their speech was not being heard or recorded, they did not entirely \*believe\* that. He went to some clever lengths to convince his 2 dozen stuttering subjects that their speech was not being heard or recorded. Result: all of his subjects were completely fluent. Something like 7 disfluencies in 10,000 syllables, which is probably better than fluent speakers. I've carried out a simple Reddit survey on this topic. My recollection is that everyone who responded did report better fluency alone than when in the presence of people, but not everyone reported full fluency when alone. Not sure how to reconcile that survey with Jackson's study, but his was a careful scientific study and mine was an informal survey. Best of luck to you going forward.