fix(flow): improve tail capture and final chunk ASR recovery

- Add FlowUtteranceEndCoordinator with 350ms silence drain and 150ms post-roll - Extend FinalChunkRecovery for short/empty final chunks in chunked pipeline - Snapshot partial at mic stop and guard final transcript in FlowSessionManager - Unify tail drain presets (iosFlow/macMLX) and expand diagnostics Co-authored-by: Rocky <hkgood@users.noreply.github.com>
2026-07-26 10:13:40 +00:00
parent e6f99d2744
commit c7ee891a90
16 changed files with 485 additions and 42 deletions
@@ -56,7 +56,11 @@ public enum UtteranceStreamChunker {
                } else if chunkIndex == 0 {
                    // Empty utterance — no chunks.
                } else {
-                    // Stream ended exactly on boundary; mark prior path complete.
+                    // Stream ended exactly on a chunk boundary; prior emit holds
+                    // all tail audio. Marker so FinalChunkRecovery paths run.
+                    continuation.yield(
+                        UtteranceAudioChunk(index: chunkIndex, samples: [], isLast: true)
+                    )
                }

                continuation.finish()