Need bg lyrics support in lrc

[MasterZack69]

I'm going to write the issue in English since I don't know chinese and I don't want any translator to misinterpret things

bg lyrics in lrc

example

[00:00.373]v1:  <00:00.373>まる <00:00.840>で <00:01.082>御伽 <00:01.565>の <00:01.733>話<00:02.771> 
[bg:[00:00.373]v1: <00:00.373>Maru <00:00.840>de <00:01.082>otogi <00:01.565>no <00:01.733>hanashi<00:02.771>]
[00:02.803]v1:  <00:02.803>終 <00:03.055>わり <00:03.370>迎え <00:03.992>た <00:04.535>証<00:04.968> 
[bg:[00:02.803]v1: <00:02.803>O<00:03.055>wari <00:03.370>mukae <00:03.992>ta <00:04.535>akashi<00:04.968>]
[00:05.136]v1:  <00:05.136>長 <00:05.379>過 <00:05.616>ぎる <00:05.994>旅路 <00:06.962>か <00:07.202>ら<00:07.851> 
[bg:[00:05.136]v1: <00:05.136>Naga<00:05.379>su<00:05.616>giru <00:05.994>tabiji <00:06.962>ka <00:07.202>ra<00:07.851>]

expected behaviour

[PhantomPixel-0418]

Hi, why do you have lyrics in LRC format where the romanized text is included in the background lyrics? (Actually, I didn’t even know LRC had a format for background lines.) Plus, there doesn’t seem to be a good way to tell whether this is actually a background line or romanized text. So, if this really is the background line format, we’ll just have to parse it as a background line—there’s no other solution.

[MasterZack69]

Hi, why do you have lyrics in LRC format where the romanized text is included in the background lyrics?

This is the only "known" way to put translation and transliteration lines in lrc, a lot of people use it for translation and some (like me) for romanized text

So, if this really is the background line format, we’ll just have to parse it as a background line—there’s no other solution.

As far as I have seen, the other music players also parse it as background line. So yes, this is the solution

[MasterZack69]

I think you could respect timestamps in background lyrics (if they exist)

[MoYingJi]

我们确实无法区分它是罗马音还是背景歌词。如果将它解析为背景歌词，它的行为会与罗马音不同。罗马音始终显示在主歌词行下方而背景歌词仅当播放到这句背景行所属的主歌词时才会显示

We genuinely cannot distinguish between Romanized text and background lyrics. If it's interpreted as background lyrics, its behavior differs from Romanized text. Romanized text always appears below the main lyrics line, while background lyrics only appear when the main lyrics corresponding to that background line are played.

图示 | Illustration

背景行仅当播放到所属的主歌词时才显示（下图标红框的位置也存在背景行，但没有显示）

The background lyrics line only appears when the corresponding main lyrics line is played (a background lyrics line also exists in the area highlighted in red in the image below, but it's not displayed).

罗马音始终显示

Romanized text always appears.

在中文社区，普遍存在的在 LRC 中标注翻译和罗马音的方法是直接将它们放在主歌词行下方，并保持与主歌词行相同的开始时间。我们之前只见过这种写法，写下了这段代码

In the Chinese community, the common method for annotating translations and Romanized lyrics in LRC is to place them directly below the main lyrics line, maintaining the same start time as the main lyrics line. We had only seen this approach before, so we wrote this code.

代码 | Code

SPlayer-Next/src/utils/lyric/parseLRC.ts

Lines 195 to 210 in 5c524eb

	// 合并同 startTime 的隔行翻译：第一行为主；第二行为 translation；第三行为 romaji；再多忽略
	const merged: LyricLine[] = [];
	for (const line of lines) {
	const prev = merged[merged.length - 1];
	if (prev && prev.startTime === line.startTime) {
	const text = line.words
	.map((w) => w.word)
	.join("")
	.trim();
	if (!text) continue;
	if (!prev.translatedLyric) prev.translatedLyric = text;
	else if (!prev.romanLyric) prev.romanLyric = text;
	continue;
	}
	merged.push(line);
	}

这种方法的前提是 LRC 不允许歌词重叠（至少不允许两行的开始时间相同），如果考虑背景行，这是允许的。如果也按这种方法解析罗马音，很可能误判真正的背景行

This method relies on LRC not allowing overlapping lyrics (at least not allowing two lines to have the same start time). If background lines are considered, this is permissible. Parsing Romanized lyrics in the same way could easily misinterpret it as a true background line.

[MasterZack69]

Can we simply parse this as however you do role="x-bg" in TTML,

the bg[] in eLRC is supposed to be the equivalent of role="x-bg"

[MoYingJi]

可以，但你得不到预期行为（如我上面所说）

Yes, but you won't get the expected behavior (as I mentioned above).

[MasterZack69]

I'm fine with that, eLRC has no single standard, it is largely defined by its community's usage. So everyone just implements them in their own way