[iOS] Dictation #2710
[iOS] Dictation #2710
Conversation
src/components/Editable.tsx
Outdated
| @@ -346,6 +346,20 @@ const Editable = ({ | |||
| ), | |||
| ) | |||
|
|
|||
| if (newValue.toLowerCase().indexOf(' new thought') > 0) { | |||
| const [prev, next] = newValue.split('new thought') | |||
There was a problem hiding this comment.
The split should also be case insensitive.
src/components/Editable.tsx
Outdated
| setTimeout(() => { | ||
| dispatch( | ||
| newThought({ | ||
| at: rootedParentOf(state, path), |
There was a problem hiding this comment.
You should omit at. It will create the new thought after the cursor thought by default. Creating the new thought at the end of the parent doesn't make sense.
src/components/Editable.tsx
Outdated
| @@ -346,6 +347,19 @@ const Editable = ({ | |||
| ), | |||
| ) | |||
|
|
|||
| if (inputMethod === 'dictation' && newValue.toLowerCase().indexOf(' new thought') > 0) { | |||
There was a problem hiding this comment.
This would result in a false positive when "new thought" is typed on the keyboard and then the user enters dictation mode. The intended behavior is for "new thought" and other commands to only be transformed when the user is dictating.
This reverts commit 4ed90f5.
When you type input event usually fired every 1-4 characters, but when you dictate whole `new thought` text appears at once
There was a problem hiding this comment.
Thanks for the update! This is an improvement, as typed "new thought" no longer triggers the dictation command. However, I'm finding that it fails to detect "new thought" in the case where dictation inserts "new" and "thought" in separate events, which is quite common. When I practiced dictating I found that I got a lot of false negatives.
I think you will need to look at the full value of the thought to pick up on these cases where "new" and "thought" are input by the dictation feature as separate words. However, you also need to avoid false positives and ignore "new thought" that was previously typed. Therefore, a better approach might be to record the selection offset when dictation starts so that you can exclude existing text from detection.
e.g. If you have thought Hello| World (with the caret in between "Hello" and "World") and you start dictating, then value.replace(/^Hello/, '').replace(/ world$/, '') will give you the dictated text excluding text that was already there.
While the expected functionality is fairly intuitive when using the app, it can be helpful to spell it out more explicitly so that you can review each of these cases and ensure that your solution works as expected:
- Trigger the New Thought command when "new thought" is dictated.
- Trigger the New Thought command when "new [pause] thought" is dictated.
- Do not trigger New Thought when "new thought" is typed.
- Do not trigger New Thought when dictating new text on a thought that already contains the text "new thought".
Thanks and let me know if I can clarify anything else!
fixes #2119