Speech-to-Text Input

Many peer mentors - particularly those serving Blindeforbundet's visually impaired user base - find keyboard entry slow or inaccessible. Allowing voice dictation after an activity lowers the barrier to completing a report, directly increasing submission rates. Higher submission rates improve data quality for Bufdir reporting and coordinator oversight. For users with motor impairments (NHF's stroke-survivor cohort), voice input can be the difference between submitting a report and abandoning it entirely. The feature also reduces the time-per-registration metric, which is the primary adoption driver identified across all workshops.

Implemented using Flutter's speech_to_text plugin, which wraps iOS SFSpeechRecognizer and Android SpeechRecognizer. Transcription is performed on-device where the platform supports it, falling back to a cloud API (with user consent) where it does not. The Speech Input Widget is a reusable widget added to any TextFormField that opts in; it manages microphone permission state, recording indicator, and interim/final result display. The widget must clearly indicate when it is listening and when it has stopped - critical for accessibility and for Blindeforbundet's requirement that users know recording is not happening during a visit. No audio is stored; only the final text transcript is retained. The feature must degrade gracefully when the device does not support speech recognition or when the user denies microphone permission.