muscle memory

i spent time in early 2024 with someone who frequently took notes on their phone using voice dictation. it was quite a thing to observe, as i’d always thought speech to text was shit. the inability to detect nuances in language, the frequent errors in or outright absence of punctuation. take a look at your average iphone voice note transcription to see what i mean.

however, this person was quite methodical with their approach. they read out the punctuations, including where they wanted to start a new paragraph, or put things in bullet points. i found it extremely fascinating - working one’s way around the obvious problem by way of adaptation.

i take a lot of notes. i am constantly pulling out my phone and typing out my thoughts, or writing out a reflection such as this one in my journal. while i admired the ability to be precise with dictation, i figured i will soon be able to leverage tooling to make this as seamless as possible for me. after all, language models had gotten really good over time and automatic speech recognition models like whisper were already available in the wild, waiting for developers to integrate into their products in functional ways. soon, i’ll finally be dictating my thoughts to my device. or so i thought.

a few weeks ago, i came across this app called monologue. it’s a mac-only voice dictation app that can be called upon when using any app. you can set up custom profiles that factor in the app or website currently in use and tailor how it should approach the transcription. pretty powerful and super configurable. i have used it a few times to much satisfaction and frankly, surprise at how well it recognizes certain words or non-english names (when i reference other people in my dictation).

yet, i don’t instinctively summon it when i have something to type. when i remember to use it, it feels awkward in some way, even though the results are nothing but satisfactory. i’ve spent over 25 years furiously being adept at using the keyboard that i have not built the muscle for voice dictation. sure there are many settings where speaking out loud isn’t ideal or could use the discretion of typing, but if we are being serious, i’m mostly at home holed up in either my study or my workshop.

i have no shortage of private space to dictate my thoughts, but unsurprisingly, here i am, fingers to keyboard, typing out this reflection.