Some people find it easy, some will find it very hard.
There may be privacy issues if people can overhear you
One of my clients likes to dictate. He sends me voice memos through messaging services. I use Whisper in Obsidian to quickly transcribe them into text, and then AI (Text Generator) to reformat/process them. Much faster than human transcription and at least as accurate.
True. Sometimes there’s nothing you can do about being overheard.
There are also privacy concerns regarding Ai transcription and data collection. It’s possible that local Ai’s could address this concern in the future.
Local AI for transcription is already available. Apps like MacWhisper use the whisper model locally. It works well, but requires a fast machine. The quality of those transcripts is amazingly good.
Overhearing is the reason why I use dictation only in some locations.
For some unclear reasons I like dictation on some devices more than others. Talking into an iPhone comes natural. Dictation on iPad or Mac feels only natural if combined with a headset. And sometimes I prefer talking into a small voice recorder without immediate transcription.
I neither prefer dictation over typing, nor typing over dictation. What works for me is a combination of both.
Typing can hurt your hands, but dictating can hurt your throat. I am not used to speak a lot, and trying to dictate a long text leave me hoarse. Also it’a pain to edit this text afterwards. That’s why I only use dictation for short notes.
On the “speaking punctuation” issue, Whisper and other AI are pretty good about putting punctuation in the right places even if you don’t dictate it. (ie you can dictate a full note without punctuation and transcribe it later with Whisper adding the punctuation in.) I know that this works well for business letters, emails, that kind of thing. Haven’t tried it for novels with dialogue.