[WIP] Audio codec probe/detection + CLI audio-info + viewer audio panel

Snapshot of local WIP before rebasing onto the movie/voice remote work.
- audio.rs: AudioCodec enum, AudioInfo::probe (RIFF/WAVE parse, raw-XMA2
  Shannon-entropy heuristic looks_like_raw_xma2), from_wav, riff_data_span.
- cli/main.rs: `audio-info` subcommand (cmd_audio_info).
- viewer ui.rs/iso_loader.rs: draw_audio_detail panel wiring.

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

crates/sylpheed-viewer/src/iso_loader.rs

@@ -165,6 +165,8 @@ pub enum PakContent {
     Ratc(Vec<RatcEntry>),
     /// Plain-text / XML payload, with its encoding label.
     Text { text: String, encoding: String },
+    /// An audio stream (WAV / XMA / raw XMA2) — metadata only; XMA isn't decoded.
+    Audio(sylpheed_formats::AudioInfo),
 }
 
 /// One child in a presented RATC bundle.
@@ -978,6 +980,12 @@ fn classify_content(payload: &[u8]) -> PakContent {
             encoding: encoding.to_string(),
         };
     }
+    // Last resort: a self-describing WAV/XMA header, or a header-less high-
+    // entropy blob (the game's raw XMA2 sound-bank streams).
+    let audio = sylpheed_formats::AudioInfo::probe(payload);
+    if audio.codec != sylpheed_formats::AudioCodec::Unknown {
+        return PakContent::Audio(audio);
+    }
     PakContent::None
 }
 

crates/sylpheed-viewer/src/ui.rs

@@ -506,6 +506,7 @@ fn draw_pak_browser(ui: &mut egui::Ui, pak: &mut PakView) {
                     PakContent::Text { text, encoding } => {
                         draw_text_detail(ui, row, text, encoding)
                     }
+                    PakContent::Audio(info) => draw_audio_detail(ui, row, info),
                     PakContent::None => draw_idxd_detail(ui, row),
                 },
             });
@@ -531,6 +532,10 @@ fn row_kind(row: &crate::iso_loader::PakRow) -> String {
         PakContent::Lsta(f) => format!("LSTA · {} sprite(s)", f.len()),
         PakContent::Ratc(c) => format!("RATC · {} item(s)", c.len()),
         PakContent::Text { .. } => "text".into(),
+        PakContent::Audio(a) => match a.xma_packets {
+            Some(p) => format!("audio · {} · {p} pkts", a.codec.label()),
+            None => format!("audio · {}", a.codec.label()),
+        },
         PakContent::None => row.identity.clone(),
     }
 }
@@ -836,6 +841,59 @@ fn draw_text_detail(
     );
 }
 
+/// An audio stream entry: codec + whatever metadata is derivable. The game's
+/// `sound.pak` streams are raw XMA2 (no embedded channels/rate), so most fields
+/// read "—" until the sound-bank descriptor + an XMA2 decoder are added.
+fn draw_audio_detail(
+    ui: &mut egui::Ui,
+    row: &crate::iso_loader::PakRow,
+    info: &sylpheed_formats::AudioInfo,
+) {
+    ui.horizontal(|ui| {
+        ui.heading("Audio stream");
+        ui.separator();
+        ui.label(info.codec.label());
+        ui.separator();
+        ui.label(format!("hash {:08x}", row.hash));
+    });
+    ui.separator();
+
+    let dash = "—".to_string();
+    egui::Grid::new("audio_meta").striped(true).num_columns(2).show(ui, |ui| {
+        ui.strong("Codec");
+        ui.label(info.codec.label());
+        ui.end_row();
+        ui.strong("Channels");
+        ui.label(info.channels.map(|c| c.to_string()).unwrap_or_else(|| dash.clone()));
+        ui.end_row();
+        ui.strong("Sample rate");
+        ui.label(info.sample_rate.map(|r| format!("{r} Hz")).unwrap_or_else(|| dash.clone()));
+        ui.end_row();
+        if let Some(d) = info.duration_secs {
+            ui.strong("Duration");
+            ui.label(format!("{d:.2} s"));
+            ui.end_row();
+        }
+        if let Some(p) = info.xma_packets {
+            ui.strong("XMA packets");
+            ui.label(format!("{p}  (2048 B each)"));
+            ui.end_row();
+        }
+        ui.strong("Size");
+        ui.label(format!("{} bytes", info.size_bytes));
+        ui.end_row();
+    });
+
+    if info.codec.needs_decoder() {
+        ui.separator();
+        ui.colored_label(
+            egui::Color32::from_gray(150),
+            "Raw XMA2 stream — playback needs an XMA2 decoder and the (un-RE'd) \
+             sound-bank descriptor for per-stream channels/sample-rate.",
+        );
+    }
+}
+
 // ── Video player (WMV cutscenes) ──────────────────────────────────────────────
 
 /// Format seconds as `mm:ss`.