diff --git a/tests/test_batch_analyze_additional.py b/tests/test_batch_analyze_additional.py
new file mode 100644
index 0000000..e2984da
--- /dev/null
+++ b/tests/test_batch_analyze_additional.py
@@ -0,0 +1,194 @@
+"""Focused tests for ``iav.batch_analyze``."""
+
+from __future__ import annotations
+
+import json
+import zipfile
+from pathlib import Path
+
+import pytest
+
+from iav.analysis_enums import MusicXmlImportMode
+from iav.batch_analyze import (
+    BatchConfig,
+    _analyze_one,
+    _load_notes_from_path,
+    run_batch,
+    run_canonical_verification,
+)
+
+
+def _batch_cfg(tmp_path: Path, **kwargs) -> BatchConfig:
+    input_dir = tmp_path / "input"
+    output_dir = tmp_path / "output"
+    input_dir.mkdir(parents=True, exist_ok=True)
+    return BatchConfig(input_dir=input_dir, output_dir=output_dir, **kwargs)
+
+
+@pytest.mark.parametrize("suffix", [".xml", ".musicxml"])
+def test_load_notes_from_path_xml_uses_aggregate_mode(tmp_path, monkeypatch, suffix):
+    cfg = _batch_cfg(tmp_path)
+    xml_path = cfg.input_dir / f"score{suffix}"
+    xml_path.write_bytes(b"<score-partwise/>")
+
+    expected_notes = [("C", 0.0, 4), ("E", 0.0, 4), ("G", 0.0, 4)]
+    captured: list[tuple[bytes, MusicXmlImportMode]] = []
+
+    def fake_parse(data, mode, **kwargs):
+        captured.append((data, mode))
+        return expected_notes, 0
+
+    monkeypatch.setattr("iav.batch_analyze.parse_musicxml_upload", fake_parse)
+
+    notes = _load_notes_from_path(xml_path, cfg)
+
+    assert notes == expected_notes
+    assert captured == [(b"<score-partwise/>", MusicXmlImportMode.AGGREGATE)]
+
+
+def test_load_notes_from_path_mxl_reads_embedded_xml(tmp_path, monkeypatch):
+    cfg = _batch_cfg(tmp_path)
+    mxl_path = cfg.input_dir / "score.mxl"
+    xml_bytes = b"<score-partwise version='3.0'/>"
+
+    with zipfile.ZipFile(mxl_path, "w") as zf:
+        zf.writestr("readme.txt", "mxl container metadata only")
+        zf.writestr("score.musicxml", xml_bytes)
+
+    captured: list[bytes] = []
+
+    def fake_parse(data, mode, **kwargs):
+        captured.append(data)
+        return [("C", 0.0, 4), ("G", 0.0, 4)], 0
+
+    monkeypatch.setattr("iav.batch_analyze.parse_musicxml_upload", fake_parse)
+
+    notes = _load_notes_from_path(mxl_path, cfg)
+
+    assert notes == [("C", 0.0, 4), ("G", 0.0, 4)]
+    assert captured == [xml_bytes]
+
+
+def test_load_notes_from_path_mxl_without_xml_raises(tmp_path):
+    cfg = _batch_cfg(tmp_path)
+    mxl_path = cfg.input_dir / "empty.mxl"
+    with zipfile.ZipFile(mxl_path, "w") as zf:
+        zf.writestr("readme.txt", "no score here")
+
+    with pytest.raises(ValueError, match="No XML inside MXL"):
+        _load_notes_from_path(mxl_path, cfg)
+
+
+def test_load_notes_from_path_unsupported_extension_raises(tmp_path):
+    cfg = _batch_cfg(tmp_path)
+    txt_path = cfg.input_dir / "notes.txt"
+    txt_path.write_text("not supported", encoding="utf-8")
+
+    with pytest.raises(ValueError, match="Unsupported input"):
+        _load_notes_from_path(txt_path, cfg)
+
+
+def test_analyze_one_fewer_than_two_notes():
+    cfg = BatchConfig(input_dir=Path("."), output_dir=Path("."))
+    row = _analyze_one("solo", [("C", 0.0, 4)], cfg)
+
+    assert row == {
+        "source_id": "solo",
+        "error": "fewer than 2 notes",
+        "note_count": 1,
+    }
+
+
+def test_analyze_one_valid_dyad_returns_metric_keys():
+    cfg = BatchConfig(input_dir=Path("."), output_dir=Path("."))
+    row = _analyze_one("dyad", [("C", 0.0, 4), ("G", 0.0, 4)], cfg)
+
+    assert row["source_id"] == "dyad"
+    assert row["note_count"] == 2
+    assert "headline_H" in row
+    assert "H_label" in row
+    assert "classification" in row
+    assert "pair_score" in row
+    assert "error" not in row
+
+
+def test_run_batch_catches_load_exception(tmp_path, monkeypatch):
+    cfg = _batch_cfg(tmp_path)
+    (cfg.input_dir / "case.json").write_text(
+        json.dumps({"notes": [["C", 0.0, 4], ["E", 0.0, 4]]}),
+        encoding="utf-8",
+    )
+
+    def boom(path, batch_cfg):
+        raise RuntimeError("load failed")
+
+    monkeypatch.setattr("iav.batch_analyze._load_notes_from_path", boom)
+
+    result = run_batch(cfg)
+
+    assert result["results"][0] == {"source_id": "case", "error": "load failed"}
+
+
+def test_run_batch_catches_processing_exception(tmp_path, monkeypatch):
+    cfg = _batch_cfg(tmp_path)
+    (cfg.input_dir / "case.json").write_text(
+        json.dumps({"notes": [["C", 0.0, 4], ["E", 0.0, 4]]}),
+        encoding="utf-8",
+    )
+
+    def boom(source_id, notes, batch_cfg):
+        raise RuntimeError("analysis exploded")
+
+    monkeypatch.setattr("iav.batch_analyze._analyze_one", boom)
+
+    result = run_batch(cfg)
+
+    assert len(result["results"]) == 1
+    assert result["results"][0]["source_id"] == "case"
+    assert result["results"][0]["error"] == "analysis exploded"
+    assert (cfg.output_dir / "results.json").is_file()
+    assert (cfg.output_dir / "results.csv").is_file()
+
+
+def test_run_batch_writes_output_files_for_success_and_error(tmp_path, monkeypatch):
+    cfg = _batch_cfg(tmp_path)
+    (cfg.input_dir / "good.json").write_text(
+        json.dumps({"notes": [["C", 0.0, 4], ["G", 0.0, 4]]}),
+        encoding="utf-8",
+    )
+    (cfg.input_dir / "bad.json").write_text(
+        json.dumps({"notes": [["C", 0.0, 4]]}),
+        encoding="utf-8",
+    )
+
+    result = run_batch(cfg)
+
+    assert len(result["results"]) == 2
+    csv_text = (cfg.output_dir / "results.csv").read_text(encoding="utf-8")
+    assert "source_id" in csv_text
+    assert "good" in csv_text
+    assert "bad" in csv_text
+    assert "fewer than 2 notes" in csv_text
+
+    saved = json.loads((cfg.output_dir / "results.json").read_text(encoding="utf-8"))
+    by_id = {row["source_id"]: row for row in saved["results"]}
+    assert by_id["bad"]["error"] == "fewer than 2 notes"
+    assert by_id["good"].get("error") is None
+
+
+def test_run_canonical_verification_success_without_output_dir():
+    assert run_canonical_verification(None) == 0
+
+
+def test_run_canonical_verification_failure_writes_report(tmp_path, monkeypatch):
+    def fail_verify(computed, expected, sonority_id, **kwargs):
+        raise AssertionError(f"metric mismatch for {sonority_id}")
+
+    monkeypatch.setattr("iav.canonical_corpus.verify_sonority_metrics", fail_verify)
+
+    assert run_canonical_verification(tmp_path) == 1
+
+    report = json.loads((tmp_path / "canonical_verification.json").read_text(encoding="utf-8"))
+    assert report["failures"]
+    assert report["results"]
+    assert all(row["ok"] is False for row in report["results"])