Skip to content

Fix Chinese lyrics detection and remove tone suffix from pinyin #2236

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
theovilardo merged 1 commit into from
Jun 4, 2026
+7 −27
Merged

Fix Chinese lyrics detection and remove tone suffix from pinyin #2236

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
34 changes: 7 additions & 27 deletions app/src/main/java/com/theveloper/pixelplay/utils/LyricsUtils.kt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -107,12 +107,12 @@ object MultiLangRomanizer {
private val KYRGYZ_SPECIFIC_CYRILLIC_LETTERS = setOf("Ң", "ң", "Ө", "ө", "Ү", "ү")
private val MACEDONIAN_SPECIFIC_CYRILLIC_LETTERS = setOf("Ѓ", "ѓ", "Ѕ", "ѕ", "Ќ", "ќ")

fun isJapanese(text: String): Boolean {
fun isJapanese(text: String, entireLyricsHasKana: Boolean = false): Boolean {
// Detect Hiragana or Katakana
if (text.any { it in '\u3040'..'\u309F' || it in '\u30A0'..'\u30FF' }) return true

// Treat Kanji-only lines as Japanese (prioritized over Chinese in the main parsing loop)
return text.any { it in '\u4E00'..'\u9FFF' }
// Treat Kanji-only lines as Japanese only if the entire song has Japanese kana
return entireLyricsHasKana && text.any { it in '\u4E00'..'\u9FFF' }
}
fun isKorean(text: String) = text.any { it in '\uAC00'..'\uD7A3' }
fun isHindi(text: String) = text.any { it in '\u0900'..'\u097F' }
Expand Down Expand Up @@ -382,28 +382,6 @@ object MultiLangRomanizer {
continue
}

// ── Tonal sandhi: 一 (yī) ──────────────────────────────────
// Before 4th-tone syllable → yí; before 1st/2nd/3rd → yì
c == '一' -> {
val nextPinyin = next?.let { getPinyin(it, format) } ?: ""
// Pinyin4j without-tone loses tone info; we approximate:
// characters that are typically 4th tone starters
val fourthToneInitials = setOf("是","不","去","做","看","用","要","过","太","对","大","会","在","上","下","个","万","次","号")
val tone = if (next != null && fourthToneInitials.contains(next.toString())) "yi2" else "yi4"
sb.append(tone.dropLast(1)).append(" ")
idx++
continue
}

// ── Tonal sandhi: 不 (bù) → bú before 4th-tone syllable ──
c == '不' -> {
val fourthToneChars = setOf("是","去","做","看","用","要","过","太","对","大","会","在","上","下","个","万","次","号","不")
val isBefore4th = next != null && fourthToneChars.contains(next.toString())
sb.append(if (isBefore4th) "bu2" else "bu4").append(" ")
idx++
continue
}

// ── Regular Hanzi ─────────────────────────────────────────
c.toString().matches(Regex("[\u4E00-\u9FA5]")) -> {
sb.append(getPinyin(c, format, prevChar = chars.getOrNull(idx - 1))).append(" ")
Expand Down Expand Up @@ -752,6 +730,8 @@ object LyricsUtils {
return Lyrics(plain = emptyList(), synced = emptyList())
}

val entireLyricsHasKana = lyricsText.any { it in '\u3040'..'\u309F' || it in '\u30A0'..'\u30FF' }

val normalizedInput = stripLeadingLyricsDocumentNoise(lyricsText)
if (looksLikeTtmlDocument(normalizedInput)) {
val converted = TtmlLyricsParser.parseToEnhancedLrc(normalizedInput)
Expand Down Expand Up @@ -888,7 +868,7 @@ object LyricsUtils {
val pairedLines = pairTranslationLines(sortedSyncedLines).map { line ->

val romanized = when {
MultiLangRomanizer.isJapanese(line.line) -> MultiLangRomanizer.romanizeJapanese(line.line)
MultiLangRomanizer.isJapanese(line.line, entireLyricsHasKana) -> MultiLangRomanizer.romanizeJapanese(line.line)
MultiLangRomanizer.isChinese(line.line) -> MultiLangRomanizer.romanizeChinese(line.line)
MultiLangRomanizer.isKorean(line.line) -> MultiLangRomanizer.romanizeKorean(line.line)
MultiLangRomanizer.isHindi(line.line) -> MultiLangRomanizer.romanizeHindi(line.line)
Expand All @@ -910,7 +890,7 @@ object LyricsUtils {
} else {
val processedPlain = plainLines.map { line ->
val romanized = when {
MultiLangRomanizer.isJapanese(line) -> MultiLangRomanizer.romanizeJapanese(line)
MultiLangRomanizer.isJapanese(line, entireLyricsHasKana) -> MultiLangRomanizer.romanizeJapanese(line)
MultiLangRomanizer.isChinese(line) -> MultiLangRomanizer.romanizeChinese(line)
MultiLangRomanizer.isKorean(line) -> MultiLangRomanizer.romanizeKorean(line)
MultiLangRomanizer.isHindi(line) -> MultiLangRomanizer.romanizeHindi(line)
Expand Down
Loading