Fastest and only macOS Dictation app with on-device STT and custom trained AI enhancement model. Windows, iOS testflight! Android, Linux waitlist open! A local Wispr Flow alternative. Follow https://x.com/fluidvoiceapp for more
perf(asr): raise Transcribing status delay to 250 ms
altic-dev526c2aa20017
diff --git a/Sources/Fluid/Services/ASRService.swift b/Sources/Fluid/Services/ASRService.swiftindex be0f970e..2eae883d 100644--- a/Sources/Fluid/Services/ASRService.swift+++ b/Sources/Fluid/Services/ASRService.swift@@ -391,7 +391,9 @@ enum ASRStopOutcome: Equatable {/// Models are cached locally to avoid repeated downloads.@MainActorfinal class ASRService: ObservableObject {- private static let finalTranscriptionStatusDelayNanoseconds: UInt64 = 100_000_000+ // Below this the Transcribing status only adds a main-thread overlay re-render+ // that delays the result it is announcing; final ASR on short audio is ~40 ms.+ private static let finalTranscriptionStatusDelayNanoseconds: UInt64 = 250_000_000private static let streamingDrainTimeoutNanoseconds: UInt64 = 30_000_000_000nonisolated static func shouldAssessShortAudioSilence(