From d4cd91f0e5d4e3c78222777caf5f8d0ece72c0a9 Mon Sep 17 00:00:00 2001
From: Michael Hansen <mike@rhasspy.org>
Date: Wed, 9 Jun 2021 13:30:34 -0400
Subject: [PATCH] Add script for re-ordering lexicon

---
 bin/reorder_lexicon.py | 73 ++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 73 insertions(+)
 create mode 100755 bin/reorder_lexicon.py

diff --git a/bin/reorder_lexicon.py b/bin/reorder_lexicon.py
new file mode 100755
index 0000000..a5387db
--- /dev/null
+++ b/bin/reorder_lexicon.py
@@ -0,0 +1,73 @@
+#!/usr/bin/env python3
+import json
+import sys
+from collections import Counter, defaultdict
+
+SKIP_WORDS = {"<eps>", "<unk>"}
+
+if len(sys.argv) < 2:
+    print(
+        "Usage: reorder_lexicon.py aligned_phones.jsonl [phoneme_col=1] < lexicon.txt > reordered_lexicon.txt"
+    )
+    sys.exit(1)
+
+aligned_path = sys.argv[1]
+phoneme_col = 1
+
+if len(sys.argv) > 2:
+    phoneme_col = int(sys.argv[2])
+
+
+lexicon = defaultdict(dict)
+num_prons = Counter()
+
+# word -> pron counts
+pron_counts = defaultdict(Counter)
+
+# Load lexicon
+print("Loading lexicon...", file=sys.stderr)
+for line in sys.stdin:
+    line = line.strip()
+    if not line:
+        continue
+
+    parts = line.split()
+    assert len(parts) > 1, line
+
+    word = parts[0]
+    phonemes = tuple(parts[phoneme_col:])
+
+    lexicon[word][phonemes] = line
+
+    num_prons[word] += 1
+
+print("Loading alignments...", file=sys.stderr)
+with open(aligned_path, "r") as aligned_file:
+    for line in aligned_file:
+        line = line.strip()
+        if not line:
+            continue
+
+        alignment = json.loads(line)
+        for word_pron in alignment["prons"]:
+            word = word_pron["word"]
+            if (word in SKIP_WORDS) or (num_prons[word] < 1):
+                continue
+
+            phonemes = tuple(word_pron["phones"])
+            pron_counts[word][phonemes] += 1
+
+
+print("Re-ordering lexicon...", file=sys.stderr)
+for word in sorted(lexicon.keys()):
+    word_pron_counts = pron_counts[word]
+    phonemes_lines = sorted(
+        lexicon[word].items(),
+        key=lambda kv: word_pron_counts.get(kv[0], 0),
+        reverse=True,
+    )
+
+    for _, line in phonemes_lines:
+        print(line)
+
+print("Done", file=sys.stderr)