Audius Lyric Transcription tasks involve listening to a music track on the Audius platform and accurately writing out the lyrics exactly as they are heard. The goal is to create a verbatim representation of all essential vocal content so the transcription reflects the actual audio performance.
These tasks help produce structured lyric datasets used for accessibility, search, and AI-related workflows. Because consistency and accuracy are essential, following formatting and transcription rules closely will reduce the chance of rejection and improve validation outcomes.
Depending on the task assignment, you may:
Listen carefully to a song hosted on Audius.
Transcribe lyrics exactly as heard.
Review or validate existing lyric submissions.
Check formatting and transcription accuracy against established standards.
Your role is to represent the performance faithfully. Do not interpret, rewrite, or improve the lyrics beyond what is present in the audio.
Use headphones when possible to improve clarity.
Work in an environment where you can listen without distractions.
Be prepared to replay difficult sections multiple times.
Follow all formatting rules carefully, even if they differ from typical lyric websites.
Your transcription must match the audio exactly for all lyrical content.
Include:
All sung lyrics
Spoken phrases that are part of the performance
Repeated sections that are clearly audible
Do not include:
Instrumental sounds
Sound effects
Non-vocal audio elements
Write out every lyric line fully, even when repeated.
Correct:
I’ll keep barking up the wrong tree
I’ll keep barking up the wrong tree
I’ll keep barking up the wrong tree
Incorrect:
I’ll keep barking up the wrong tree
(Repeat 3x)
Repeated lyrics must always be written exactly as performed.
Consistency in formatting helps validation and improves dataset quality.
Treat each lyric line like a written sentence.
Capitalize the first word of each line.
Capitalize proper nouns, names, locations, and acronyms.
Do not write entire lines in ALL CAPS.
Use minimal punctuation.
Do not add periods or commas at the end of lines.
Question marks, exclamation points, and quotation marks are allowed when clearly expressed in the audio.
Avoid repeated punctuation for emphasis.
Punctuation inside a line is allowed when grammatically necessary.
Non-word vocals such as “ooh,” “ah,” or “mmm” may be included if they meaningfully contribute to the song.
Use parentheses for background vocals.
Correct:
(You're a black room boy)
Incorrect:
(door slamming)
Do not include environmental sounds or instrumental noises.
If a track includes overlapping or sampled vocals containing clear lyrical content, include them in the transcription.
Do not transcribe:
Purely instrumental samples
Sound effects
Explicit language should be transcribed exactly as heard.
If audio censors part of a word, replace missing letters with asterisks.
Example:
Audio: “I’m a f(break)r”
Lyrics: I’m a f****r
Do not censor or modify words unless the audio itself is censored.
Validators should focus on accuracy and adherence to formatting rules.
Check for:
Exact matching between audio and transcription
Missing or added words
Proper handling of repeated lines
Correct capitalization and punctuation
Validation prioritizes faithful transcription over stylistic preference. For a handy "Validator Cheat Sheet," click here.
Using shorthand such as “repeat x2” instead of writing lyrics fully.
Adding labels like “[Verse]” or “[Chorus]”.
Including musical notation or structural commentary.
Leaving out lightly sung or spoken words that are clearly audible.
Guessing lyrics without careful listening.
Accurate lyric transcription ensures:
Lyrics reflect the actual performance.
Validation and review systems operate consistently.
Data remains reliable for downstream workflows.
Contributors are rewarded fairly for high-quality submissions.
Careful attention to these guidelines helps maintain consistency and trust across the platform.