ConvAgent: add waveform-splice discontinuity debug detector

Diagnostic for the audio "pops": EnqueueAgentAudio compares the last sample of
the previous enqueued block with the first sample of the new one and logs
[AUDIO-DBG] with the delta + a running count when the jump exceeds a threshold
(gated behind ps.ai.ConvAgent.Debug.ElevenLabs; role-tagged). The host's raw-PCM
stream is continuous so its count stays ~0; a client whose count climbs at
message cadence still has a splice problem. Confirms the carry-over fix
(bcb6c44) at runtime.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

Unreal/PS_AI_Agent/Plugins/PS_AI_ConvAgent/Source/PS_AI_ConvAgent/Private/PS_AI_ConvAgent_ElevenLabsComponent.cpp

@@ -1248,12 +1248,26 @@ void UPS_AI_ConvAgent_ElevenLabsComponent::HandleAudioReceived(const TArray<uint
 				* static_cast<int32>(sizeof(int16));            // 640 at 16kHz mono
 			static constexpr int32 MaxOpusChunkBytes = 40 * BytesPerFrame; // 25600 = 0.8s
 
-			// Frame-align the message length (Encode drops any sub-frame remainder
+			// Carry the sub-frame remainder from the PREVIOUS message so no samples
-			// anyway). A frame-aligned length guarantees every sub-chunk holds >=1
+			// are dropped at message boundaries. Opus encodes whole 640-byte frames
-			// whole frame and lets us reliably mark the FINAL sub-chunk (bLast), so
+			// only; dropping each message's <1-frame tail left a ~10-16ms hole and a
-			// the client reassembles one message into a single enqueue (no per-chunk
+			// waveform discontinuity at every message splice → audible pops on clients
-			// playback re-arm → no pops).
+			// (the host plays raw PCM directly, so standalone/host never had it).
-			const int32 UsableBytes = (PCMData.Num() / BytesPerFrame) * BytesPerFrame;
+			// Prepend the leftover, encode only whole frames, stash the new tail.
+			if (!bAgentSpeaking)
+			{
+				// First message of a new turn — don't glue on the previous turn's tail.
+				AgentAudioEncodeLeftover.Reset();
+			}
+			TArray<uint8> ContinuousPCM = MoveTemp(AgentAudioEncodeLeftover);
+			ContinuousPCM.Append(PCMData);
+
+			const int32 UsableBytes = (ContinuousPCM.Num() / BytesPerFrame) * BytesPerFrame;
+			if (ContinuousPCM.Num() > UsableBytes)
+			{
+				AgentAudioEncodeLeftover.Append(
+					ContinuousPCM.GetData() + UsableBytes, ContinuousPCM.Num() - UsableBytes);
+			}
 
 			int32 Offset = 0;
 			while (Offset < UsableBytes)
@@ -1262,7 +1276,7 @@ void UPS_AI_ConvAgent_ElevenLabsComponent::HandleAudioReceived(const TArray<uint
 				const bool bLast = (Offset + BlockBytes >= UsableBytes);
 
 				uint32 CompressedSize = static_cast<uint32>(OpusWorkBuffer.Num());
-				OpusEncoder->Encode(PCMData.GetData() + Offset, BlockBytes,
+				OpusEncoder->Encode(ContinuousPCM.GetData() + Offset, BlockBytes,
 					OpusWorkBuffer.GetData(), CompressedSize);
 
 				if (CompressedSize > 0)
@@ -1718,6 +1732,42 @@ void UPS_AI_ConvAgent_ElevenLabsComponent::OnProceduralUnderflow(
 
 void UPS_AI_ConvAgent_ElevenLabsComponent::EnqueueAgentAudio(const TArray<uint8>& PCMData)
 {
+	// Debug: flag a waveform discontinuity at the splice between the previous
+	// enqueued block and this one — the signature of the audio pops. The host's
+	// raw-PCM stream is continuous so its count should stay ~0; if a client's
+	// count climbs at message cadence, a splice problem remains.
+	{
+		const int32 DbgVal = CVarDebugElevenLabs.GetValueOnGameThread();
+		const bool bAudioDbg = (DbgVal >= 0) ? (DbgVal > 0) : bDebug;
+		if (bAudioDbg && PCMData.Num() >= static_cast<int32>(sizeof(int16)))
+		{
+			const int16* Samples = reinterpret_cast<const int16*>(PCMData.GetData());
+			const int32 NumSamples = PCMData.Num() / static_cast<int32>(sizeof(int16));
+			if (bHasLastEnqueuedSample)
+			{
+				const int32 Delta = FMath::Abs(static_cast<int32>(Samples[0]) - static_cast<int32>(LastEnqueuedSample));
+				static constexpr int32 DiscontinuityThreshold = 3000; // |delta| on the int16 scale
+				if (Delta > DiscontinuityThreshold)
+				{
+					++AudioSpliceDiscontinuityCount;
+					UE_LOG(LogPS_AI_ConvAgent_ElevenLabs, Warning,
+						TEXT("[AUDIO-DBG] Role=%d splice discontinuity: |delta|=%d (prev=%d first=%d) count=%d — pop likely here."),
+						static_cast<int32>(GetOwnerRole()), Delta,
+						static_cast<int32>(LastEnqueuedSample), static_cast<int32>(Samples[0]),
+						AudioSpliceDiscontinuityCount);
+				}
+				else if (DbgVal >= 2)
+				{
+					UE_LOG(LogPS_AI_ConvAgent_ElevenLabs, Log,
+						TEXT("[AUDIO-DBG] Role=%d splice ok: |delta|=%d"),
+						static_cast<int32>(GetOwnerRole()), Delta);
+				}
+			}
+			LastEnqueuedSample = Samples[NumSamples - 1];
+			bHasLastEnqueuedSample = true;
+		}
+	}
+
 	{
 		FScopeLock Lock(&AudioQueueLock);
 		AudioQueue.Append(PCMData);
@@ -1892,7 +1942,9 @@ void UPS_AI_ConvAgent_ElevenLabsComponent::StopAgentAudio()
 	bool bWasSpeaking = false;
 	double Now = 0.0;
 	bPreBuffering = false;      // Clear pre-buffer state on stop.
-	ClientReassemblyPCM.Reset(); // Drop any half-received message so it can't leak into the next turn.
+	ClientReassemblyPCM.Reset();       // Client: drop any half-received message so it can't leak into the next turn.
+	AgentAudioEncodeLeftover.Reset();  // Server: drop the carried sub-frame tail so it can't glue onto the next turn.
+	bHasLastEnqueuedSample = false;    // Debug: don't count the inter-turn splice as a discontinuity.
 	{
 		FScopeLock Lock(&AudioQueueLock);
 		AudioQueue.Empty();

Unreal/PS_AI_Agent/Plugins/PS_AI_ConvAgent/Source/PS_AI_ConvAgent/Public/PS_AI_ConvAgent_ElevenLabsComponent.h

@@ -812,6 +812,21 @@ private:
 	// sub-chunk (bLastSubChunk). Game-thread only. Reset in StopAgentAudio().
 	TArray<uint8> ClientReassemblyPCM;
 
+	// Server-only: sub-frame (<640 byte) tail carried from the previous agent
+	// audio message. Opus encodes whole frames only; without carry-over each
+	// message dropped its <1-frame tail, leaving a ~10-16ms hole + waveform
+	// discontinuity at every message splice → audible pops on clients. Reset
+	// per turn (new-turn guard in HandleAudioReceived + StopAgentAudio()).
+	TArray<uint8> AgentAudioEncodeLeftover;
+
+	// Debug: detects waveform discontinuities at enqueue/message splices (the
+	// signature of the audio "pops"). Tracks the last enqueued sample; a large
+	// jump to the next block's first sample is flagged and counted. Toggle with
+	// ps.ai.ConvAgent.Debug.ElevenLabs. Reset in StopAgentAudio().
+	int16 LastEnqueuedSample = 0;
+	bool  bHasLastEnqueuedSample = false;
+	int32 AudioSpliceDiscontinuityCount = 0;
+
 	// Pre-buffer state: delay playback start to absorb TTS inter-chunk gaps.
 	bool bPreBuffering = false;
 	double PreBufferStartTime = 0.0;