batchalign 0.8.2.post8__tar.gz → 0.8.2.post10__tar.gz

{batchalign-0.8.2.post8/batchalign.egg-info → batchalign-0.8.2.post10}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: batchalign
-Version: 0.8.2.post8
+Version: 0.8.2.post10
 Summary: Python Speech Language Sample Analysis
 Author: Brian MacWhinney, Houjun Liu
 Author-email: macw@cmu.edu, houjun@cmu.edu

{batchalign-0.8.2.post8 → batchalign-0.8.2.post10}/batchalign/pipelines/analysis/compare.py

@@ -216,6 +216,30 @@ def _find_best_segment(gold_tokens, main_tokens, mfn):
     return best
 
 
+def _best_rotation(window_tokens, gold_tokens, mfn):
+    """Find the cyclic rotation of *window_tokens* that maximises matches.
+
+    Returns the rotation offset *r* such that
+    ``window_tokens[r:] + window_tokens[:r]`` best aligns to *gold_tokens*.
+    """
+    if len(window_tokens) <= 1:
+        return 0
+
+    best_r = 0
+    best_matches = -1
+    n = len(window_tokens)
+
+    for r in range(n):
+        rotated = window_tokens[r:] + window_tokens[:r]
+        alignment = align(rotated, gold_tokens, False, mfn)
+        matches = sum(1 for item in alignment if isinstance(item, Match))
+        if matches > best_matches:
+            best_matches = matches
+            best_r = r
+
+    return best_r
+
+
 def _get_pos(form):
     """Extract uppercased POS from a Form's morphology, or '?' if absent."""
     if form is not None and form.morphology:
@@ -287,6 +311,8 @@ class CompareEngine(BatchalignEngine):
             for form_idx, form in enumerate(utt.content):
                 if form.text.strip() in MOR_PUNCT + ENDING_PUNCT:
                     continue
+                if _get_pos(form) == "PUNCT":
+                    continue
                 if form.text.strip().lower() in fillers:
                     continue
                 main_info.append((utt_idx, form_idx, form))
@@ -303,7 +329,7 @@ class CompareEngine(BatchalignEngine):
         for utt_idx, utt in enumerate(gold_utterances):
             gold_punct[utt_idx] = []
             for form_idx, form in enumerate(utt.content):
-                if form.text.strip() in MOR_PUNCT + ENDING_PUNCT:
+                if form.text.strip() in MOR_PUNCT + ENDING_PUNCT or _get_pos(form) == "PUNCT":
                     gold_punct[utt_idx].append((form_idx, form))
                     continue
                 if form.text.strip().lower() in fillers:
@@ -360,8 +386,13 @@ class CompareEngine(BatchalignEngine):
                     utt_main_forms[utt_idx].append(m_form)
                     utt_main_speakers[utt_idx].append(m_utt_idx)
 
-            # Align the chosen window against this gold utterance
+            # Align the chosen window against this gold utterance,
+            # trying cyclic rotations to avoid spurious del/ins pairs.
             window_main = conformed_main[abs_start:abs_end]
+            window_len = len(window_main)
+            rotation = _best_rotation(window_main, g_tokens, match_fn)
+            if rotation > 0:
+                window_main = window_main[rotation:] + window_main[:rotation]
             utt_alignment = align(window_main, g_tokens, False, match_fn)
 
             local_main_cursor = 0
@@ -370,7 +401,7 @@ class CompareEngine(BatchalignEngine):
 
             for item in utt_alignment:
                 if isinstance(item, Match):
-                    global_main_idx = abs_start + local_main_cursor
+                    global_main_idx = abs_start + (local_main_cursor + rotation) % window_len
                     orig_main_idx = main_map[global_main_idx]
                     main_form = main_info[orig_main_idx][2]
                     orig_gold_idx = g_maps[local_gold_cursor]
@@ -401,7 +432,7 @@ class CompareEngine(BatchalignEngine):
                         local_gold_cursor += 1
 
                     else:
-                        global_main_idx = abs_start + local_main_cursor
+                        global_main_idx = abs_start + (local_main_cursor + rotation) % window_len
                         orig_main_idx = main_map[global_main_idx]
                         main_form = main_info[orig_main_idx][2]
 
@@ -462,6 +493,10 @@ class CompareEngine(BatchalignEngine):
                 utt.time = (timed_forms[0].time[0], timed_forms[-1].time[1])
                 utt.text = None
 
+        # Copy @Media header from the main doc if it has one
+        if doc.media is not None:
+            gold.media = doc.media
+
         return gold
 
 

{batchalign-0.8.2.post8 → batchalign-0.8.2.post10}/batchalign/version

@@ -1,3 +1,3 @@
-0.8.2-post.8
+0.8.2-post.10
 April 02 2026
 segment

{batchalign-0.8.2.post8 → batchalign-0.8.2.post10/batchalign.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: batchalign
-Version: 0.8.2.post8
+Version: 0.8.2.post10
 Summary: Python Speech Language Sample Analysis
 Author: Brian MacWhinney, Houjun Liu
 Author-email: macw@cmu.edu, houjun@cmu.edu