As a Swiftkey Customer I have two major areas where I think the developers could improve upon.
The first one is voice to text.
1) While I like the option to translate voice to text it feels somewhat unrefined. One issue is that it often stops listening before I've finished speaking. I find myself speaking rapid fire with no interruptions in order to keep swiftkey listening.
2) While it's nice to have the 3 alternate options, in practice, it takes time to sort through the differences, find the version with the closest accuracy and select it. One suggestion would be to highlight in a different color the words that are in common and the words that are different.
3) Voice to Text accuracy overall is somewhat low. I seem to have better luck with Vlingo although this is still grading on a curve as no solution is 100% accurate. It seems like voice to text as a concept toes the fine line between being a time saver and being more trouble than it's worth.
The second one is Auto-Complete/Auto-Accept.
1) I'm not sure what the difference here is actually. Both descriptions sound like it does the same thing.
2) Here's my take, it's both amazing in it's corrective capability and sometimes infuriating when it replaces correctly spelled words with it's own suggestions, multiple times!! I think if the user spells a word once and has auto-complete on then it should replace the word only if it's not spelled correctly. If I change it back 2 times, it should leave the word alone even if it's not spelled correctly.
3) Again this is a fantastic feature that spells and auto-predicts the next logical word but if I end up with a text that makes no sense because I'm not policing the auto-complete vigilantly, it can feel somewhat foolish.
4) Email addresses. Swiftkey doesn't learn entire email address as one string but breaks it up into Joeblow @ gmail.com. This makes it frustrating to enter as you have to go back and delete the spaces that it puts between the parts of the address.
I hope these comments are taken under consideration...
The first one is voice to text.
1) While I like the option to translate voice to text it feels somewhat unrefined. One issue is that it often stops listening before I've finished speaking. I find myself speaking rapid fire with no interruptions in order to keep swiftkey listening.
2) While it's nice to have the 3 alternate options, in practice, it takes time to sort through the differences, find the version with the closest accuracy and select it. One suggestion would be to highlight in a different color the words that are in common and the words that are different.
3) Voice to Text accuracy overall is somewhat low. I seem to have better luck with Vlingo although this is still grading on a curve as no solution is 100% accurate. It seems like voice to text as a concept toes the fine line between being a time saver and being more trouble than it's worth.
The second one is Auto-Complete/Auto-Accept.
1) I'm not sure what the difference here is actually. Both descriptions sound like it does the same thing.
2) Here's my take, it's both amazing in it's corrective capability and sometimes infuriating when it replaces correctly spelled words with it's own suggestions, multiple times!! I think if the user spells a word once and has auto-complete on then it should replace the word only if it's not spelled correctly. If I change it back 2 times, it should leave the word alone even if it's not spelled correctly.
3) Again this is a fantastic feature that spells and auto-predicts the next logical word but if I end up with a text that makes no sense because I'm not policing the auto-complete vigilantly, it can feel somewhat foolish.
4) Email addresses. Swiftkey doesn't learn entire email address as one string but breaks it up into Joeblow @ gmail.com. This makes it frustrating to enter as you have to go back and delete the spaces that it puts between the parts of the address.
I hope these comments are taken under consideration...