batchalign 0.7.19.post10__tar.gz → 0.7.19.post12__tar.gz

{batchalign-0.7.19.post10/batchalign.egg-info → batchalign-0.7.19.post12}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: batchalign
-Version: 0.7.19.post10
+Version: 0.7.19.post12
 Summary: Python Speech Language Sample Analysis
 Author: Brian MacWhinney, Houjun Liu
 Author-email: macw@cmu.edu, houjun@cmu.edu

{batchalign-0.7.19.post10 → batchalign-0.7.19.post12}/batchalign/pipelines/analysis/eval.py

@@ -38,9 +38,47 @@ def conform(x):
         elif "wanna" == i.strip():
             result.append("want")
             result.append("to")
+        elif "ii" == i.strip():
+            result.append("i")
+            result.append("i")
+        elif "i'd" == i.strip():
+            result.append("i")
+            result.append("had")
+        elif "tshirts" == i.strip():
+            result.append("t")
+            result.append("shirts")
+        elif "tshirts" == i.strip():
+            result.append("t")
+            result.append("shirts")
+        elif "anytime" == i.strip():
+            result.append("any")
+            result.append("time")
+        elif "alright" == i.strip():
+            result.append("all")
+            result.append("right")
+        elif "sorta" == i.strip():
+            result.append("sort")
+            result.append("of")
+        elif "alrightie" == i.strip():
+            result.append("all")
+            result.append("right")
+        elif "mm" == i.strip():
+            result.append("hm")
+        elif "ai" == i.strip():
+            result.append("a")
+            result.append("i")
+        elif "this'll" == i.strip():
+            result.append("this")
+            result.append("will")
         elif "gotta" == i.strip():
             result.append("got")
             result.append("to")
+        elif "eh" == i.strip():
+            result.append("uh")
+        elif "kinda" == i.strip():
+            result.append("a")
+            result.append("kind")
+            result.append("of")
         elif "farmhouse" == i.strip():
             result.append("farm")
             result.append("house")
@@ -57,6 +95,11 @@ def conform(x):
 
     return result
 
+def match_fn(x,y):
+    return (y == x or
+            y.replace("(", "").replace(")", "") == x.replace("(", "").replace(")", "") or
+            re.sub(r"\((.*)\)",r"", y) == x or re.sub(r"\((.*)\)",r"", x) == y)
+
 class EvaluationEngine(BatchalignEngine):
     tasks = [ Task.WER ]
 
@@ -69,8 +112,8 @@ class EvaluationEngine(BatchalignEngine):
         forms = [i.replace("-", "") for i in forms if i.strip() not in MOR_PUNCT+ENDING_PUNCT]
         gold_forms = [i.replace("-", "") for i in gold_forms if i.strip() not in MOR_PUNCT+ENDING_PUNCT]
 
-        forms = [re.sub(r"\((.*)\)",r"", i) for i in forms]
-        gold_forms = [re.sub(r"\((.*)\)",r"", i) for i in gold_forms]
+        # forms = [re.sub(r"\((.*)\)",r"", i) for i in forms]
+        # gold_forms = [re.sub(r"\((.*)\)",r"", i) for i in gold_forms]
 
         # if there are single letter frames, we combine them tofgether
         # until the utterance is done or there isn't any left
@@ -111,7 +154,7 @@ class EvaluationEngine(BatchalignEngine):
         forms_final = conform(forms_final)
 
         # dp!
-        alignment = align(forms_final, gold_final, False)
+        alignment = align(forms_final, gold_final, False, match_fn)
 
         # calculate each type of error
         sub = 0

{batchalign-0.7.19.post10 → batchalign-0.7.19.post12}/batchalign/utils/dp.py

@@ -76,7 +76,7 @@ def __serialize_arr(src, tgt):
 
     return src_serialized, tgt_serialized
 
-def __dp(payload, reference, t):
+def __dp(payload, reference, t, match_fn):
     """Performs bottom-up dynamic programming alignment
 
     Parameters
@@ -149,7 +149,7 @@ def __dp(payload, reference, t):
             # get a match.
 
             # recall 1 indexing
-            is_match = (reference[i-1].key == payload[j-1].key)
+            is_match = match_fn(reference[i-1].key, payload[j-1].key)
 
             # calculate new distances
             new_dist1 = dist1+(0 if is_match else 2)
@@ -209,15 +209,16 @@ def __dp(payload, reference, t):
 
 def align(source_payload_sequence,
           target_reference_sequence,
-          tqdm=True):
+          tqdm=True,
+          match_fn=lambda x,y: x==y):
     """Align two sequences"""
 
     if (len(source_payload_sequence) > 0 and
         type(source_payload_sequence[0]) == PayloadTarget):
-        return __dp(source_payload_sequence, target_reference_sequence, tqdm)
+        return __dp(source_payload_sequence, target_reference_sequence, tqdm, match_fn)
     else:
         return __dp(*__serialize_arr(source_payload_sequence,
-                                     target_reference_sequence), tqdm)
+                                     target_reference_sequence), tqdm, match_fn)
 
 # align([1,2,3,4,4,5,5,5], [1,1,3,4,4,12,5,5,18])
 

batchalign-0.7.19.post12/batchalign/version

@@ -0,0 +1,3 @@
+0.7.19-post.12
+July 10st, 2025
+Fix ASR evals paren matchin

{batchalign-0.7.19.post10 → batchalign-0.7.19.post12/batchalign.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: batchalign
-Version: 0.7.19.post10
+Version: 0.7.19.post12
 Summary: Python Speech Language Sample Analysis
 Author: Brian MacWhinney, Houjun Liu
 Author-email: macw@cmu.edu, houjun@cmu.edu

batchalign-0.7.19.post10/batchalign/version

@@ -1,3 +0,0 @@
-0.7.19-post.10
-July 1st, 2025
-Whoops, fixed ASR regression.