DartBridgeSTT class
STT component bridge for C++ interop.
Provides thread-safe access to the C++ STT component. Handles model loading, transcription, and streaming.
Properties
- currentModelId → String?
-
Get the currently loaded model ID.
no setter
- hashCode → int
-
The hash code for this object.
no setterinherited
- isLoaded → bool
-
Check if a model is loaded.
no setter
- runtimeType → Type
-
A representation of the runtime type of the object.
no setterinherited
- supportsStreaming → bool
-
Check if streaming is supported.
no setter
Methods
-
destroy(
) → void - Destroy the component and release resources.
-
getHandle(
) → RacHandle - Get or create the STT component handle.
-
loadModelForStreaming(
{required String path, required String id, required String name}) → void -
Load a model onto this component handle. Streaming sessions are
handle-bound, so the lifecycle-resolved model must be loaded here first
(mirrors Swift
prepareStreamingHandle). No-op when already loaded. -
noSuchMethod(
Invocation invocation) → dynamic -
Invoked when a nonexistent method or property is accessed.
inherited
-
toString(
) → String -
A string representation of this object.
inherited
-
transcribeLifecycleProto(
STTTranscriptionRequest request) → STTOutput - Transcribe audio through the lifecycle-owned generated-proto STT ABI.
-
transcribeLifecycleProtoAsync(
STTTranscriptionRequest request) → Future< STTOutput> -
Transcribe audio through the lifecycle-owned generated-proto STT ABI,
running the blocking native call in a short-lived worker isolate
(
Isolate.run) so the calling isolate — usually the Flutter UI isolate — stays responsive for the whole transcription. Whisper decode of a batch buffer is a long synchronous block; on the UI isolate it freezes frames. Mirrorsdart_bridge_llm.dart'sgenerateProto: this ABI is lifecycle-owned (no Dart-held handle), so the worker re-resolves the engine via the commons model lifecycle — nothing isolate-bound crosses. -
transcribeProto(
Uint8List audioData, STTOptions options) → Future< STTOutput> - Transcribe audio with serialized runanywhere.v1.STTOptions.
-
transcribeSessionStream(
Stream< Uint8List> audio, STTOptions options) → Stream<STTPartialResult> - Canonical chunk-feed stream-in / stream-out transcription session.
Operators
-
operator ==(
Object other) → bool -
The equality operator.
inherited
Static Properties
-
Shared instance
final
Static Methods
-
setTranscribeLifecycleProtoForTesting(
STTOutput override(STTTranscriptionRequest)?) → void