Adding weighted WER feature.

Author: RuABraun

.github/workflows/publish-to-pypi.yaml

@@ -8,7 +8,7 @@ jobs:
     if: startsWith(github.ref, 'refs/tags')
     strategy:
       matrix:
-        os: [ubuntu-latest, ubuntu-24.04, macos-13, macos-14, windows-latest, macos-latest]
+        os: [ubuntu-latest, macos-13, macos-14, windows-latest]
 
     steps:
       - uses: actions/checkout@v4
@@ -51,7 +51,7 @@ jobs:
     strategy:
       matrix:
         python-version: ["3.10", "3.12"]
-        os: [ubuntu-latest, ubuntu-24.04, macos-13, macos-14, windows-latest, macos-latest]
+        os: [ubuntu-latest, macos-13, macos-14, windows-latest]
 
     steps:
       - uses: actions/checkout@v4
@@ -91,6 +91,11 @@ jobs:
           path: dist
           merge-multiple: true
 
+      - name: Check dist files
+        run: |
+          ls -lh dist/
+          file dist/*
+
       - uses: pypa/[email protected]
         with:
           user: __token__

MANIFEST.in

@@ -1,2 +1,3 @@
 include LICENSE README.md requirements.txt
 recursive-include libs *.*
+recursive-include texterrors/data *

README.md

@@ -10,6 +10,7 @@ Features:
 - Metrics by group (for example speaker)
 - Comparing two hypothesis files to reference
 - Oracle WER
+- **NEW** Weighted WER (English only)
 - Sorting most common errors by frequency or count
 - Measuring performance on keywords
 - Measuring OOV-CER (see [https://arxiv.org/abs/2107.08091](https://arxiv.org/abs/2107.08091) )
@@ -89,6 +90,7 @@ This results in a WER of 83.3\% because of the extra insertion and deletion. And
 
 Recent changes:  
 
+- 11.11.25 Weighted WER for English
 - 26.02.25 Faster alignment, better multihyp support, fixed multihyp bug.
 - 22.06.22 refactored internals to make them simpler, character aware alignment is off by default, added more explanations
 - 20.05.22 fixed bug missing regex dependency

requirements.txt

@@ -6,3 +6,4 @@ termcolor
 Levenshtein
 regex
 pytest
+importlib_resources

setup.py

@@ -4,7 +4,7 @@
 import setuptools
 import sys
 
-__version__ = "1.0.10"
+__version__ = "1.0.12"
 
 
 class get_pybind_include(object):
@@ -101,5 +101,7 @@ def get_requires():
     entry_points={'console_scripts': ['texterrors=texterrors.texterrors:cli']},
     install_requires=get_requires(),
     setup_requires=['pybind11'],
-    python_requires='>=3.6'
+    python_requires='>=3.6',
+    package_data={"texterrors": ["data/wordlist"]},
+    include_package_data=True,
 )

tests/test_functions.py

@@ -117,6 +117,21 @@ def test_oov_cer():
     assert err / cnt == 0., err / cnt
 
 
+def test_weighted_wer():
+    reflines = ['1 my name is john doe']
+    hyplines = ['1 my name is joe doe']
+    refs = create_inp(reflines)
+    hyps = create_inp(hyplines)
+    buffer = io.StringIO()
+    texterrors.process_output(refs, hyps, buffer, 'A', 'B',weighted_wer=True, skip_detailed=True)
+    output = buffer.getvalue()
+    ref ="""WER: 20.0 (ins 0, del 0, sub 1 / 5)
+SER: 100.0
+Weighted WER: 23.1
+"""
+    assert output == ref, show_diff(output, ref)
+
+
 def test_seq_distance():
     a, b = 'a b', 'a b'
     d = texterrors.seq_distance(StringVector(a.split()), StringVector(b.split()))