I was not aware of any speech recognition tools that I have used that did not use machine learning. It’s been around for quite some time (>20 years). The Wikipedia page lists some systems from before the 1980s that didn’t use machine learning though I’m unsure whether those were ever used.