When you use Adobe Podcast to enhance the quality of audio recordings, and there is noise that is not someone speaking, it wants to force the generated audio into words. There is definitely a fair amount of rebuilding happening in the process, but it's otherwise so opaque without any controls or configuration options.