fixing alignments

lhotse/supervision.py

@@ -671,7 +671,12 @@ def with_alignment_from_ctm(
                     num_overspanned += len(alignment)
                     segments.append(fastcopy(seg, alignment={type: alignment}))
             else:
-                segments.append([s for s in self.find(recording_id=reco_id)])
+                segments.extend(
+                    [
+                        fastcopy(s, alignment={type: []})
+                        for s in self.find(recording_id=reco_id)
+                    ]
+                )
         logging.info(
             f"{num_overspanned} alignments added out of {num_total} total. If there are several"
             " missing, there could be a mismatch problem."

test/fixtures/supervision.json

@@ -14,5 +14,17 @@
     "recording_id": "recording-1",
     "start": 0.5,
     "duration": 0.4
+  },
+  {
+    "id": "segment-3",
+    "recording_id": "recording-2",
+    "start": 0.1,
+    "duration": 0.2
+  },
+  {
+    "id": "segment-4",
+    "recording_id": "recording-2",
+    "start": 0.3,
+    "duration": 0.2
   }
 ]

test/test_supervision_set.py

@@ -180,6 +180,10 @@ def test_supervision_set_with_alignment_from_ctm(
 ):
     segment = external_supervision_set["segment-1"]
     assert external_alignment == segment.alignment
+    assert external_supervision_set["segment-2"].alignment == {"word": []}
+    assert external_supervision_set["segment-3"].alignment == {"word": []}
+    for seg in external_supervision_set:
+        assert type(seg) == SupervisionSegment
 
 
 def test_supervision_set_write_alignment_to_ctm(external_supervision_set, tmp_path):