diff --git a/packages/frontend/src/scripts/check-word-mute.ts b/packages/frontend/src/scripts/check-word-mute.ts
index 67e896b4b..6f3c6c40d 100644
--- a/packages/frontend/src/scripts/check-word-mute.ts
+++ b/packages/frontend/src/scripts/check-word-mute.ts
@@ -3,40 +3,63 @@
  * SPDX-License-Identifier: AGPL-3.0-only
  */
 
-export function checkWordMute(note: Record<string, any>, me: Record<string, any> | null | undefined, mutedWords: Array<string | string[]>): boolean {
+import type { Note, MeDetailed } from "misskey-js/entities.js";
+
+// TODO: this implementation is horribly inefficient.
+// Each regex is validated (using a regex, ironically), transformed, and then parsed - for each note being checked.
+// These regex objects should be cached somewhere.
+
+export function checkWordMute(note: Note, me: MeDetailed | null | undefined, mutedWords: Array<string | string[]>): boolean {
 	// 自分自身
 	if (me && (note.userId === me.id)) return false;
 
-	if (mutedWords.length > 0) {
-		const text = ((note.cw ?? '') + '\n' + (note.text ?? '')).trim();
+	if (mutedWords.length < 1) return false;
 
-		if (text === '') return false;
+	const text = getNoteText(note);
+	if (text === '') return false;
 
-		const matched = mutedWords.some(filter => {
-			if (Array.isArray(filter)) {
-				// Clean up
-				const filteredFilter = filter.filter(keyword => keyword !== '');
-				if (filteredFilter.length === 0) return false;
+	return mutedWords.some(filter => {
+		if (Array.isArray(filter)) {
+			// Clean up
+			const filteredFilter = filter.filter(keyword => keyword !== '');
+			if (filteredFilter.length === 0) return false;
 
-				return filteredFilter.every(keyword => text.includes(keyword));
-			} else {
-				// represents RegExp
-				const regexp = filter.match(/^\/(.+)\/(.*)$/);
+			return filteredFilter.every(keyword => text.includes(keyword));
+		} else {
+			// represents RegExp
+			const regexp = filter.match(/^\/(.+)\/(.*)$/);
 
+			// This should never happen due to input sanitisation.
+			if (!regexp) return false;
+
+			try {
+				return new RegExp(regexp[1], regexp[2]).test(text);
+			} catch (err) {
 				// This should never happen due to input sanitisation.
-				if (!regexp) return false;
-
-				try {
-					return new RegExp(regexp[1], regexp[2]).test(text);
-				} catch (err) {
-					// This should never happen due to input sanitisation.
-					return false;
-				}
+				return false;
 			}
-		});
-
-		if (matched) return true;
-	}
-
-	return false;
+		}
+	});
+}
+
+function getNoteText(note: Note): string {
+	const textParts: string[] = [];
+
+	if (note.cw)
+		textParts.push(note.cw);
+
+	if (note.text)
+		textParts.push(note.text);
+
+	if (note.files)
+		for (const file of note.files)
+			if (file.comment)
+				textParts.push(file.comment);
+
+	if (note.poll)
+		for (const choice of note.poll.choices)
+			if (choice.text)
+				textParts.push(choice.text);
+
+	return textParts.join('\n').trim();
 }