🏠 Overview › 🧩 Components ›🧩 SpeechToTextService

SpeechToTextService

Component Detail

Service Layer medium complexity mobile
0
Dependencies
0
Dependents
0
Entities
0
Integrations

Description

Wraps the device's native speech recognition API and manages the dictation session lifecycle. It normalises transcription results across platforms and returns a final transcript string for review. Errors from the underlying API are translated into user-friendly states.

Feature: Speech-to-Text Input

speech-to-text-service

Sources & reasoning

Lines 86-88 provide the precise use-case constraint (post-interaction dictation, not live recording). Line 347 places it in Fase 2, which maps to v1.0 by ordinal. Two organizations (Blindeforbundet + HLF) endorsed it, making it a "should have" cross-org feature.

  • Begge organisasjoner ønsker mulighet for ĂĄ snakke inn rapporter fremfor ĂĄ skrive. Blindeforbundet understreker at opptak under selve samtalen er uønsket - tale-til-tekst er for etterpĂĄ, ved rapportskriving.
  • Tale-til-tekst

Responsibilities

  • Initialise and tear down native speech recognition sessions
  • Stream partial transcript results to callers during active dictation
  • Return final confirmed transcript on session end
  • Normalise punctuation and casing in raw transcription output
  • Map native API errors to typed application error codes

Interfaces

startSession(locale: string): SpeechSession
stopSession(sessionId: string): Promise<TranscriptResult>
getPartialTranscript(sessionId: string): string
cancelSession(sessionId: string): void
isAvailable(): boolean