Transcription Workflow
A transcription is the written record of what can actually be heard in one specific version of a song. Its purpose is accuracy, not copying whichever lyric page appears first.
1. Identify the exact recording
Make sure the audio you are transcribing is the same version that will use the TTML. Radio edits, clean versions, remasters, deluxe releases, live recordings, sped-up versions, and regional releases may have different words, censorship, intros, or timing.
Use the song link supplied with the submission and compare other versions only as references.
2. Listen before trusting a source
Existing lyrics can save time, but they are evidence—not the authority. Genius, Musixmatch, Spotify, Apple Music, official booklets, videos, and artist posts can all contain omissions or errors.
Start with the most complete source you can find.
Check every line against the exact recording.
Restore repeats that a source abbreviates.
Correct words that disagree with the audio.
Check whether an “official” source belongs to a different version.
Official booklets and artist-provided lyrics are useful for names, unusual spellings, and difficult phrases, but the performed recording still wins when they differ.
3. Transcribe the whole performance
Include every intelligible sung or spoken word:
lead lyrics;
repeated verses and choruses;
ad-libs and background vocals;
spoken introductions, dialogue samples, producer tags, radio clips, and crowd chants;
clearly performed vocalizations.
Do not add text that is not heard. Section labels such as [Verse], [Chorus], and [Instrumental], stage directions, descriptions of sounds, and manually written instrumental ellipses are not lyrics.
4. Resolve uncertain words
Replay the surrounding phrase, slow the audio without changing its pitch, compare other releases or live performances, and isolate the vocal when helpful. Context, rhyme, grammar, and the artist’s usual vocabulary can narrow the possibilities, but they do not justify a guess.
Never submit [?], guessed filler, or another placeholder. If a lyric cannot yet be established confidently, the transcription is not finished.
AI-generated transcription is not allowed. Audio separation may be used as a listening aid, but a person must determine and verify the words.
5. Perform a final lyric-only pass
Before timing, read while listening from beginning to end and check:
every word is present and in the right order;
repeats have been written at every occurrence;
ad-libs, speech, and background vocals were not skipped;
punctuation and line breaks reflect the performance;
censorship matches this exact version;
the main lyrics use the original language and writing system.
Then continue with Spelling Standards.