From ad7be62321f5f0df6ccc20b63b9315f6b9251aa1 Mon Sep 17 00:00:00 2001
From: Dipshet <264011288+Dipshet@users.noreply.github.com>
Date: Wed, 5 Aug 2026 20:26:02 +0200
Subject: [PATCH] apu: decode XMA frames whose 15-bit header straddles a packet
 boundary

The frame walk treated fewer than 15 remaining bits as end-of-chain, but a frame starting there is real: its header continues in the next packet, as the encoder's own seek ledger confirms. Each one was silently dropped, so a premix's parallel streams drifted apart in 512-sample steps and doubled on the fold. Cvar audio_xma_header_straddle_fix, on.
---
 .../src/native/audio/audio_system.cpp         | 14 +++++++++
 .../src/native/audio/xma/context.cpp          | 30 +++++++++++++++++++
 2 files changed, 44 insertions(+)

diff --git a/thirdparty/rexglue-sdk/src/native/audio/audio_system.cpp b/thirdparty/rexglue-sdk/src/native/audio/audio_system.cpp
index df138624..1b230216 100644
--- a/thirdparty/rexglue-sdk/src/native/audio/audio_system.cpp
+++ b/thirdparty/rexglue-sdk/src/native/audio/audio_system.cpp
@@ -70,6 +70,20 @@ REXCVAR_DEFINE_BOOL(audio_xma_loop_guard, true, "Audio",
                     "machinery on every input buffer swap, truncating/"
                     "skipping subframes of the stream (confirmed firing on "
                     "AC6 cutscene streams via the deep-trace counter).");
+REXCVAR_DEFINE_BOOL(audio_xma_header_straddle_fix, true, "Audio",
+                    "Decode XMA frames whose 15-bit frame header straddles a "
+                    "2 KiB packet boundary. The frame walk treated "
+                    "fewer-than-15 remaining bits as end-of-chain, but the "
+                    "straddled frame is real: the encoder's own seek ledger "
+                    "confirms it on every occurrence (its size field, read "
+                    "across the boundary, equals the next packet's "
+                    "first-frame offset minus the packet header). Without "
+                    "this, one whole 512-sample frame is silently dropped per "
+                    "occurrence, so the parallel streams of a multi-stream "
+                    "voice drift apart in 512-sample steps and the premix "
+                    "combs on the stereo fold - the long-reported doubled/"
+                    "robotic cutscene dialogue. Off restores the legacy "
+                    "walk for A/B.");
 REXCVAR_DEFINE_BOOL(audio_xma_preserve_timeline, true, "Audio",
                     "When the XMA decoder backend rejects a damaged frame, emit that "
                     "frame's worth of silence instead of dropping it. The input read "
diff --git a/thirdparty/rexglue-sdk/src/native/audio/xma/context.cpp b/thirdparty/rexglue-sdk/src/native/audio/xma/context.cpp
index cb39582a..c9fc0a38 100644
--- a/thirdparty/rexglue-sdk/src/native/audio/xma/context.cpp
+++ b/thirdparty/rexglue-sdk/src/native/audio/xma/context.cpp
@@ -19,6 +19,7 @@
 REXCVAR_DECLARE(bool, audio_deep_trace);
 REXCVAR_DECLARE(bool, audio_xma_loop_guard);
 REXCVAR_DECLARE(bool, audio_xma_preserve_timeline);
+REXCVAR_DECLARE(bool, audio_xma_header_straddle_fix);
 
 namespace rex::audio {
 
@@ -517,6 +518,35 @@ kPacketInfo XmaContext::GetPacketInfo(uint8_t* packet, uint32_t frame_offset) {
 
   while (true) {
     if (stream.BitsRemaining() < kBitsPerFrameHeader) {
+      // A frame whose 15-bit header STRADDLES the packet boundary is real.
+      // The chain only reaches this point when the previous frame's
+      // continuation bit promised another frame (or the packet's declared
+      // first-frame offset points here), and the encoder's own seek ledger
+      // confirms it: the straddled size field, read across the boundary,
+      // equals the next packet's first_frame_offset minus the 32-bit packet
+      // header on every occurrence in AC6's demopacks. Treating <15
+      // remaining bits as end-of-chain mislabelled the previous frame "last
+      // in packet", so Decode()'s advance jumped to the next packet's
+      // first_frame_offset - exactly PAST the straddled frame's continuation
+      // - and one whole 512-sample frame vanished silently (~1 per stream
+      // per 14 s in AC6 cutscene premixes, at stream-specific positions).
+      // The streams of a multi-stream voice drifted apart in 512-sample
+      // steps and the coherent premix combed on the stereo fold: the
+      // long-reported robotic cutscene echo.
+      //
+      // Counting the frame is the whole fix: current_frame_size_ = 0 routes
+      // Decode() into its existing cross-packet path, whose combined
+      // two-payload bitstream reads the straddled header and copies the full
+      // frame. (BitsRemaining() == 0 exactly - a header starting precisely
+      // at the payload junction - keeps the old behaviour; no occurrence
+      // exists in the packs and the read offset cannot represent it.)
+      if (REXCVAR_GET(audio_xma_header_straddle_fix) && stream.BitsRemaining() > 0) {
+        if (stream.offset_bits() == frame_offset) {
+          packet_info.current_frame_ = packet_info.frame_count_;
+          packet_info.current_frame_size_ = 0;
+        }
+        packet_info.frame_count_++;
+      }
       break;
     }