YuE2INDEPENDENT AI MUSIC GENERATOR
Lyrics in.
A song out.
Turn original lyrics and a focused style brief into a complete song. Switch to music-to-music when an owned reference recording should guide the phrasing or pace of the next take.
Independent of WaveSpeed. Check the workspace availability label before generating.
Text to music
Music to music
Lyrics required
MP3 download
STARTING POINTS
Each example pairs a focused sound direction with lyrics that use section tags. Load one into the workspace, then replace the sample words with lyrics you own.
Bedroom folk, tender male vocal, fingerpicked guitar, brushed drums, quiet verse into a memorable chorus.
[Verse] We packed the morning in a paper cup Drove until the radio gave up [Chorus] If the road still knows my name Sing it back the long way home
Glossy synth-pop, bright duet, analog arpeggios, punchy bass, 118 BPM, late-night energy.
[Verse] City glass, a quieter heart Counting lights that never start [Chorus] Stay in the chorus, don’t fade out Let the night keep us loud
Cinematic instrumental bed, restrained strings, slow percussion build, spacious mix, and a wide final chorus.
[Instrumental] [Verse] A whispered line against the strings [Chorus] Hold the horizon, then let it ring
MEET YUE2
This lyrics-first studio turns a style brief and structured song words into a complete track. You can start from text alone or let one owned reference recording guide a new rendition, then keep the finished audio in your private library.
FROM DRAFT TO TRACK
Separate the songwriting decisions from the production decisions. Clear lyrics tell the model what happens in the song; a concise style brief tells it how the song should feel.
Start with one genre and one mood, then add tempo, instruments, vocal character, language, and production texture. Prioritize the details that listeners should notice first.
Paste original words with section tags and intentional line breaks. Keep the hook concise, vary the verses, and use a bridge only when the song needs a clear change of perspective.
Stay with text-to-music for a clean start. Switch routes and upload one audio file only when an owned performance should influence the phrasing, pace, or overall musical character.
Listen to the full result before changing the brief. If the direction is close, revise one variable at a time—tempo, instrumentation, vocal tone, or lyric structure—so the next result has a clear purpose.
PRACTICAL USES
AI-generated music is most useful when each take answers a specific creative question. Decide what you are testing before you generate, then evaluate the result against that goal.
Turn a lyric draft into a listenable sketch before booking a vocalist or producer. A demo can reveal whether the chorus arrives soon enough, whether a verse is too dense, and whether the emotional arc survives outside the page. Treat the output as a composition aid: note what works, rewrite weak lines, and generate a new take from the revised song.
Test the same words against contrasting production directions—a sparse acoustic performance, a bright electronic arrangement, or a cinematic build. Keep the lyrics fixed while changing one musical variable at a time. This makes the comparison useful and helps you decide which tempo, instrumentation, and vocal character best serve the hook.
Use an owned rehearsal, voice memo, or instrumental sketch when timing and phrasing matter more than a purely written brief can express. Add the new lyric structure and describe the intended production separately. The goal is a fresh rendition guided by musical information you are allowed to submit, not a duplicate of a released recording.
BEFORE YOU GENERATE
Use original or properly licensed material, review the credit quote before submitting, and keep the first prompt focused enough that revisions remain meaningful.
YuE2 is a family of lyrics-to-song models. This independent studio provides access to the 3B model through WaveSpeed inference and is not an official product page for either company.
It is a 3-billion-parameter music model that can work from lyrics and a style description, with an additional route for reference-guided generation. This site connects to the two corresponding WaveSpeed endpoints.
Lead with genre and mood, then add tempo, key instruments, vocal character, language, and production texture. Put song words in the lyrics field rather than mixing them into the style description.
No. You provide the lyrics. Draft original words first, label the sections, and keep instrumental or production direction in the style field instead of leaving the lyrics field empty.
It combines one owned reference recording with your lyrics and style brief to create a new rendition. Use a source you have permission to upload, and treat it as guidance rather than a request for an exact copy.
Choose a clean recording that communicates the rhythm, phrasing, or musical feel you want to preserve. Avoid noisy mixes or files containing material you do not have permission to process.
Change one dimension at a time. Tighten the chorus, clarify the tempo, remove conflicting genres, or name the leading instrument. Focused revisions make it easier to understand what improved the next take.
The workspace quotes credits before submission. Those quotes are derived from WaveSpeed list prices of $0.10 for text-to-music and $0.15 for music-to-music, plus the site safety margin.
No. yue2.pro is an independent YuE2 AI music generator. WaveSpeed and YuE2 names belong to their respective owners.