I used Willow Voice daily as my main voice keyboard for a few months. Then I switched to Wispr Flow, and it's what I use now and what I recommend on this site. Here's why, and how the two compare.
What I put through Willow before switching:
Willow Voice does one thing I could not turn off: it rewrites what I say. The clearest case is the word "delete". When I dictate an edit, an email, an app screen, and I want to say "delete that sentence" or "delete from here" as literal text, Willow reads "delete" as an instruction. Instead of leaving my words as I said them, it tries to edit the whole paragraph I just spoke, as if I'd told it to. What comes back is a rephrase that rarely makes sense, and the words I actually said are gone.
There is no setting to switch this off. For dictating changes to text, which is half of what I use a voice keyboard for, it was a dealbreaker.
Wispr Flow does the job I want from a voice keyboard: it types what I say. So far that's meant roughly 90% less rewriting than I had with Willow. I say the words, the words land. When I dictate an edit that contains "delete", it types "delete".
Accuracy has been higher for me too, fewer wrong words and fewer passes to fix transcription mistakes. Speed feels the same; I haven't timed it, but nothing about it drags.
The pieces I relied on are there. Wispr has a macOS app alongside the iOS one, so dictation works on the desktop as well as the phone. Double-tapping the fn key locks dictation on, so I can keep talking without holding a key down.
The onboarding is heavier than it needs to be. More setup steps before you're actually dictating than I'd like to sit through. Once you're past it, it stays out of the way, which is the part that matters day to day.
ShellDrop's SSH input field takes text from any iOS keyboard, voice keyboards included. I dictate commands and messages to a Claude Code agent running in tmux on my Mac Mini instead of thumb-typing them on a phone. The v4.0 write-up covers how that works. The voice keyboard I reach for when I do it is Wispr Flow.
If you want to try it, here's my Wispr Flow referral link.
What they do with your words. Willow Voice rewrites and reinterprets what I say, and treats a spoken "delete" as an edit instruction with no way to turn it off. Wispr Flow transcribes what I say. That's why I switched.
In my use so far it types what I say, with about 90% less rewriting than I had on Willow. Dictate an edit containing "delete" and it types "delete".
Yes. There's a macOS app alongside the iOS one. Double-tapping the fn key locks dictation on so you can keep talking hands-free.
It's the one I moved to after using Willow daily for a few months. For me: more accurate, far less rewriting, about the same speed. The trade-off is a heavier onboarding flow. Willow still has cloud sync and a custom dictionary.
Yes. A voice keyboard feeds text into any standard iOS text input, and ShellDrop's SSH input field is one. You can dictate a command or a message to a Claude Code or Codex agent in tmux instead of typing it on a phone.
ShellDrop is on the App Store. Universal iPhone and iPad, iOS 17+.
