feat: persist voice samples to disk, generate VTT, cleanup after analysis

internal/domain/port/speech_transcriber.go

@@ -2,6 +2,12 @@ package port
 
 import "context"
 
-type SpeechTranscriber interface {
-	Transcribe(ctx context.Context, audioData []byte, mimeType string) (string, error)
+// TranscriptionResult holds both the plain-text transcript and the WebVTT subtitle content.
+type TranscriptionResult struct {
+	Text string
+	VTT  string // WebVTT format, empty if not available
+}
+
+type SpeechTranscriber interface {
+	Transcribe(ctx context.Context, audioData []byte, mimeType string) (TranscriptionResult, error)
 }