Author: Yaki
Men really can’t resist these strange and unusual things.
It started when a colleague began using Doubao Input Method last year. By holding down the spacebar, you could directly convert speech into text, with a large language model correcting the wording. I even downloaded it specifically to try it out at the time. But because I was used to using the spacebar to control cursor movement, losing that function felt particularly uncomfortable. In the end, I had no choice but to give up and continue using WeChat Input Method.
Although WeChat’s voice input can only remain active for 30 minutes each time it is opened, after 30 minutes, tapping speech-to-text redirects you to the WeChat Input Method app once, and then back to WeChat. It is a little annoying, but at least I could retain control of the cursor, and the clipboard is shared between the Mac and iPhone. In the end, I gave up on that input method, and the matter came to an abrupt end.
My interest was rekindled
A few days ago, I saw someone sharing how they used a DJI microphone with a computer, paired it with two instances of Doubao’s model through Lightning Talk, and used voice input directly instead of typing. I suddenly became interested—my computer setup has three screens and two operating systems (Mac / Windows), each running different tools and working on multiple things concurrently. Sometimes I type until my keyboard runs out of battery. If I could use voice input, it should make things a little easier, right?
I started experimenting with it, and after a lot of tinkering, it really did reduce some of the frequent "frustration" of typing. At first, I still wasn’t used to it and continued typing, but I gradually began to adapt.
An unexpected discovery: voice exposed a weakness in my communication
As I increased the frequency of using voice input, I began thinking about a new question.
I had always thought that I expressed myself fairly clearly and that my written descriptions were reasonably coherent, but as soon as I switched to voice input, I began stumbling and becoming disorganized.
I fell into deep thought—as I communicated more with AI, I could clearly feel that my ability to express myself had become more precise, whether in terms of definitions, descriptions, or the accuracy of events. To make the answers returned by AI relatively accurate, I had successfullybeen domesticated by AI. I gradually adapted to communicating with AI; when typing, I would delete what I had written and type it again, then delete it and retype it, all to describe things clearly.
But voice input is completely different—when typing, you can still make corrections, but with voice input there’s basically no way to correct yourself. You can only slow down your speech, thinking about the structure of what you want to say while also thinking about how to say it.
This suddenly made me realize:There’s a huge mountain between describing something clearly by typing and speaking clearly the moment you open your mouth—not only is it heavy, it’s all uphill.
I originally thought people from Northeast China had an advantage when it came to language, but now it seems that’s not the case when it comes to expressing themselves clearly.
Deliberate Practice
At that moment, I remembered something a gaming streamer once said:
"If you feel like you’ve hit a bottleneck, start speeding up—speed up your thinking and speed up your output."
I started deliberately requiring myself to use voice input to train my thinking speed and speaking accuracy. After a day of converting voice to text, I feel like my ability to express myself has improved just a tiny bit.
Let’s keep it up and let voice input train mybrain and mouth to work as onethen! Hahahahahaha
Bonus: Instantly defeated by the official version
Also: WeChat on Mac added a voice input feature today. Press and hold Fn and it brings up voice recognition. It’s even faster at transcribing than the large model I built myself, and it also has a correction function. I was instantly defeated!!
Hahaha, although I was instantly defeated, at least I don’t have to use up any TOKENs anymore. That still feels pretty good~
I’m Yaki. Good night 🌙