How to Get Word-by-Word Lyrics on Apple Music: A Delivery Guide for Artists and Labels

Some songs on Apple Music light up word by word as they're sung. Others jump a whole line at a time, and some just show a static page of lyrics. The difference comes down to the lyrics file delivered to Apple: whether it has timing at all, and how fine that timing is.

This guide is for artists, labels and distributors who want word-by-word lyrics on their own releases. It covers what separates the three kinds of lyrics, who can actually deliver word-level timing to Apple, the rules your file has to follow, and how to time lyrics you already have without tapping out every word by hand.

Three kinds of lyrics, one difference: timing

StaticLine by lineWord by word

II keptkept thethe wordswords II nevernever saidsaid

foldedfolded inin thethe eveningevening windwind

waitingwaiting byby youryour windowwindow

toto readread themthem asas youyou passpass

I kept the words I never said / folded in the evening wind / waiting by your window / to read them as you pass

Word-by-word lyrics are the "real-time lyrics" behind Apple Music Sing. Apple's instruction to content providers is short: to be considered for beat-by-beat lyrics display, deliver lyrics in the updated TTML format (Apple Music Sing delivery notes).

Word-by-word lyrics are a TTML file

TTML is a timed-text format that looks a bit like HTML. Here's one line of a lyrics file in the shape Apple expects:

<tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en">
  <body>
    <div>
      <p begin="0:12.400" end="0:15.100">
        <span begin="0:12.400" end="0:12.650">I</span>
        <span begin="0:12.650" end="0:13.000">kept</span>
        <span begin="0:13.000" end="0:13.200">the</span>
        <!-- and so on for the rest of the line -->
      </p>
    </div>
  </body>
</tt>

• Each section of the song is a <div> and each line is a <p>. Timing on the <p> only gives you line-by-line lyrics.
• Wrap the words inside a <p> in <span>s with their own timing and you get word by word. A span can be a word, a syllable or a short phrase.
• For the whole song to count as word by word, every line needs timed spans. If one line is missing them, the file can only be delivered as line by line.

The details live in the TTML chapter of Apple's Video and Audio Asset Guide.

Who can deliver word-by-word lyrics to Apple

So typing lyrics into a dashboard, or using a distributor's tap-along tool, gets you line-by-line lyrics at best.

To find a distributor that handles lyrics, start with Apple's own Apple Music Partner Directory. It lists Apple's preferred distributors, lets you filter by region, and shows which ones offer lyrics. That column doesn't separate plain, line-level and word-level lyrics, though, so ask your shortlist two questions: Can you deliver word-level lyrics as TTML? Can I supply the TTML file myself?

Delivering doesn't guarantee display

Apple's lyrics guidelines are blunt about it: time-synced lyrics are implemented at Apple's editorial discretion (Apple lyrics guidelines). A word-level TTML file gets your song considered; it doesn't promise anything.

In my experience delivering releases with plain-text lyrics, most of those songs only ever show static lyrics on Apple Music. Apple doesn't fill in word timing for them. If word-by-word display matters to you, you have to make the timing and deliver it yourself.

One more thing: bad word timing is worse than none. Words that light up half a beat early or late feel wrong immediately. And once Apple has verified your lyrics, redelivering them won't update them (Apple: Submit lyrics). Watch and listen to the whole song before the first delivery.

Pre-delivery checklist

• Time against the exact audio you deliver. Remixes, instrumentals and extended versions each need their own timing.
• Deliver at least 5 days before release so lyrics are live on release day.
• Write out every repeat. No "x2", no "[Chorus]".
• No periods or commas at the end of a line. Question marks and exclamation points are fine.
• Background vocals go in parentheses. In TTML, mark them as background vocals with ttm:role="x-bg".
• Use the right language code in xml:lang. For Simplified Chinese lyrics, Apple accepts cmn-Hans in my experience delivering Chinese releases; Apple's public docs don't spell this out.
• One singer's lines can't overlap. If the next line starts before the previous one ends, the file breaks Apple's format rules.

The first five come from Apple's lyrics guidelines and Submit lyrics.

How to do it: lines first, then words

Timing every word by hand takes a long time. It's much easier in three passes:

1. Get every line in place. Give an alignment tool the delivery audio and the lyrics you already have. It times each line automatically, and you spot-check a few.
2. Watch the word-by-word effect and fix what feels off. Listen while the words light up, and drag the boundary between two words where the highlight lags or rushes.
3. Export word-level TTML and send it to your distributor.

That's the workflow LyricSmith is built around. Paste your lyrics when you upload the audio; the AI only adds timing and never rewrites your words. After alignment, switch to "Word by word" and watch the words light up the way they do on Apple Music. When you're done, pick "TTML word by word" in the export menu. If you edited a line's text after alignment, its word timing has to be re-estimated, and the editor flags it so you can check. If you only need line sync, pick "TTML line by line". It's free and there's no sign-up.

Try it with your song →

AI alignment isn't perfect: long intros, wordless vocals and repeated choruses can push a line or two out of place. Whatever tool you use, listen through the whole song before you deliver.