Voice input is a convenient mode that enables users to overcome literacy and/or accessibility barriers because it is typically easier than typing and enables hands-free use of the device. On devices such as smartphones or tablets, a virtual keyboard with voice input capability is convenient and enables voice input to be provided to any application. However, virtual keyboard voice interfaces lack editing capabilities, often requiring users to start from scratch multiple times because of input mistakes, transcription errors, or automatic turning off of the microphone during pauses. This disclosure describes techniques that automatically provide users with suggestions for edits, deletions, and next words during voice interaction with a virtual keyboard. With user permission, the suggestions can be generated using suitable speech and language models to analyze the user’s speech, keeping the microphone on until the user ends. The proposed suggestions can be presented in a visual interface with any suitable interactive mechanisms that enable the user to incorporate or dismiss the suggestions.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.