You Can See What You're Saying Now
Live transcription overlay for YEN's Speech-to-Text.
Hey folks!
Small feature, big difference in feel. When you hold **Option + Space** to dictate, you can now see the words appearing in real-time as you speak.
A translucent pill floats at the top of your screen, showing exactly what Apple’s speech engine is hearing. Speak, watch the words form, release, done. No more wondering “is it getting this right?” while staring at a blinking menu bar icon.
The Problem Was Uncertainty
YEN has had speech-to-text since v0.58. Hold Option + Space, speak, release, text appears. It works. But there was always this moment of faith between speaking and releasing the key. You couldn’t see anything. Just a blinking dot in the menu bar.
For short commands — “git commit minus m fix the thing” — that’s fine. You release, glance at the output, move on.
But for longer dictation? Paragraphs? Commit messages? You’re speaking into the void for 30 seconds, hoping the speech engine caught everything, and you don’t find out until you release the key and text dumps into your cursor position.
The irony is that the partial results were already streaming. Every word, every correction, every mid-sentence revision — all of it was flowing through DictationRecognizer in real-time.
The callback was wired. The data was there. It was going nowhere.
A Floating Pill
When you hold Option + Space now, a small translucent capsule appears at the top-center of your screen, right below the menu bar. It says “Listening...” until you start speaking, then it fills with your words in real-time.
The pill uses Apple’s .ultraThinMaterial — the same frosted glass effect you see in macOS system UI. A red microphone icon sits on the left. Text flows in from the right. If you’re verbose (and honestly, I always am), the text truncates from the left so you always see the most recent words.
When you release the key, the pill fades out and text pastes at your cursor. Same behavior as before, just with 100% more visibility into what’s happening.
You’ll never dictate blind again.
— 8