Catches more verbatim differences, in time for Interspeech 2024 itself

revdotcom · Aug 28, 2024 · 677be8e · 677be8e
1 parent 4d87f36
commit 677be8e
Show file tree

Hide file tree

Showing 7 changed files with 45,555 additions and 41,835 deletions.
diff --git a/coraal-multi/code/Analysis.ipynb b/coraal-multi/code/Analysis.ipynb
@@ -187,7 +187,11 @@
     "#Only added some alternatives, though these are never looked for with the functions, only added them for potential future reference\n",
     "\n",
     "#CORAAL filler words\n",
-    "fillers = {\"uh-huh\":[\"uh huh\"], \"uh-uh\":[\"uh uh\",\"nuh-uh\"],\"nuh-uh\":[\"nuh uh\",\"nuh-huh\",\"uh-uh\"], \"mm-hm\":[\"mm hm\",\"mm hmm\"], \"mm\":[\"hm\"], \n",
+    "#fillers = {\"uh-huh\":[\"uh huh\"], \"uh-uh\":[\"uh uh\",\"nuh-uh\"],\"nuh-uh\":[\"nuh uh\",\"nuh-huh\",\"uh-uh\"], \"mm-hm\":[\"mm hm\",\"mm hmm\"], \"mm\":[\"hm\"], \n",
+    "           #\"mm-mm\":[\"mm mm\"], \"okay\":[\"ok\", \"kay\"],\"mkay\":[\"mm kay\"], \"yep\":[\"yup\"], \"yup\":[\"yep\"], \"nah\":[\"naw\"],\"naw\":[\"nah\"],\"ooh\":[\"oo\"]}\n",
+    "#Above dictionary resulted in the figures submitted for Interspeech 2024, below dictionary catches more verbatim differences\n",
+    "#The new figures from the below dictionary support the conclusions we drew in that paper more strongly than the original figures\n",
+    "fillers = {\"uh-huh\":[\"uh huh\"], \"uh-uh\":[\"uh uh\",\"nuh-uh\"],\"nuh-uh\":[\"nuh uh\",\"nuh-huh\",\"uh-uh\"], \"mm-hm\":[\"mm hm\",\"mm hmm\",\"mm-hmm\"], \"mm\":[\"hm\"], \n",
     "           \"mm-mm\":[\"mm mm\"], \"okay\":[\"ok\", \"kay\"],\"mkay\":[\"mm kay\"], \"yep\":[\"yup\"], \"yup\":[\"yep\"], \"nah\":[\"naw\"],\"naw\":[\"nah\"],\"ooh\":[\"oo\"]}\n",
     "filler_no_alts = [\"oh\",\"ayo\",\"hoo\",\"uh\",\"um\",\"ah\"]\n",
     "\n",

diff --git a/coraal-multi/code/Visualizations.ipynb b/coraal-multi/code/Visualizations.ipynb
diff --git a/coraal-multi/csvs/categorized_errors.csv b/coraal-multi/csvs/categorized_errors.csv
diff --git a/coraal-multi/csvs/error_sources_summary.csv b/coraal-multi/csvs/error_sources_summary.csv
diff --git a/coraal-multi/visualizations/ASRvHumanTranscriptsWER.png b/coraal-multi/visualizations/ASRvHumanTranscriptsWER.png
diff --git a/coraal-multi/visualizations/SourceHypPercents.png b/coraal-multi/visualizations/SourceHypPercents.png
diff --git a/coraal-multi/visualizations/VerbatimHeatMap.png b/coraal-multi/visualizations/VerbatimHeatMap.png