batchalign 0.7.10__tar.gz → 0.7.10.post0__tar.gz

{batchalign-0.7.10/batchalign.egg-info → batchalign-0.7.10.post0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: batchalign
-Version: 0.7.10
+Version: 0.7.10.post0
 Summary: Python Speech Language Sample Analysis
 Author: Brian MacWhinney, Houjun Liu
 Author-email: macw@cmu.edu, houjun@cmu.edu

{batchalign-0.7.10 → batchalign-0.7.10.post0}/batchalign/pipelines/analysis/eval.py

@@ -37,7 +37,7 @@ class EvaluationEngine(BatchalignEngine):
         is_single = False
 
         for i in forms:
-            if len(i) == 1:
+            if len(i) == 1 and ("zho" not in doc.langs):
                 single_sticky += i
             else:
                 if single_sticky != "":
@@ -49,8 +49,23 @@ class EvaluationEngine(BatchalignEngine):
             forms_finished.append(single_sticky)
             single_sticky = ""
 
+        # special Chinese processing
+        gold_final = []
+        forms_final = []
+
+        if "zho" in doc.langs:
+            for i in gold_forms:
+                for j in i:
+                    gold_final.append(j)
+            for i in forms_finished:
+                for j in i:
+                    forms_final.append(j)
+        else:
+            gold_final = gold_forms
+            forms_final = forms_finished
+
         # dp!
-        alignment = align(forms_finished, gold_forms, False)
+        alignment = align(forms_final, gold_final, False)
 
         # calculate each type of error
         sub = 0

{batchalign-0.7.10 → batchalign-0.7.10.post0}/batchalign/pipelines/asr/utils.py

@@ -27,7 +27,7 @@ def retokenize(intermediate_output):
             word = word.replace("。", ".")
             word = word.replace("¿", " ").replace("¡", " ")
             tmp.append((word, bullet))
-            if word in ENDING_PUNCT or word[-1] in ENDING_PUNCT:
+            if len(word) > 0 and (word in ENDING_PUNCT or word[-1] in ENDING_PUNCT):
                 if word in ENDING_PUNCT:
                     final_outputs.append((speaker, tmp))
                 elif word[-1] in ENDING_PUNCT:
@@ -39,7 +39,7 @@ def retokenize(intermediate_output):
                 tmp = []
 
         if len(tmp) > 0: 
-            if tmp[-1][0] in MOR_PUNCT:
+            if len(tmp[-1]) > 0 and tmp[-1][0] in MOR_PUNCT:
                 tmp.pop(-1)
             tmp.append((".", [None, None]))
             final_outputs.append((speaker, tmp))

batchalign-0.7.10.post0/batchalign/version

@@ -0,0 +1,3 @@
+0.7.10-post.0
+Janurary 22th, 2025
+some align fixes

{batchalign-0.7.10 → batchalign-0.7.10.post0/batchalign.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: batchalign
-Version: 0.7.10
+Version: 0.7.10.post0
 Summary: Python Speech Language Sample Analysis
 Author: Brian MacWhinney, Houjun Liu
 Author-email: macw@cmu.edu, houjun@cmu.edu

batchalign-0.7.10/batchalign/version

@@ -1,3 +0,0 @@
-0.7.10
-Janurary 20th, 2025
-insert options multi automatically