Say what the song should feel like. The AI writes the words, the music and sings it.
AI music generator
ImgAI creates AI music from a description or your own lyrics, with vocal and instrumental options. Describe the mood, genre and instruments, compare the generated versions, then use the music tools to extend a song, try another arrangement or prepare files for editing.
Create a song in three steps
- 01
Describe the music or add lyrics
Write a mood, genre and instrumentation in Describe mode. Choose Write lyrics to provide your own words and style, or turn on Instrumental for a track without vocals.
- 02
Check the settings and credits
Sign in, choose a model and review the cost before submitting. Target length is a request, not an exact output duration. Max mode and follow-up tools can change the credit cost.
- 03
Listen, compare and continue
Find the generated tracks in your workspace. Listen to each version, check lyrics and transitions, then download a track or choose an editing action from its menu.
Continue working with your music
Extend or rearrange a song
Continue from a chosen point, request a cover in another style or use two songs as references for a mashup. These actions generate new results; voices, lyrics and arrangements may change.
Work with vocals and stems
Request vocals, drums, bass or another available stem from an existing song. To add vocals or accompaniment to your own recording, import the audio through the matching tool first. Review the separation and mix before using the result.
Prepare audio, lyrics and notes
Download a generated track or request MP3, WAV or M4A conversion. When lyric alignment is available, export an LRC file with line timestamps. MIDI transcription is a separate request; a downloadable file depends on the returned result.
Music tools
AI music generation FAQ
Yes. Write lyrics mode accepts your lyrics and style instructions. Describe mode starts from a written idea. Turn on Instrumental for music without a vocal part. Listen to the result to check the wording, pronunciation and arrangement.
A standard generation typically returns two versions for comparison. The workspace shows the tracks actually returned. Extensions, stem separation and other actions have different result types, so two tracks are not promised for every operation.
Music generation requires an account and credits. The lyrics-writing tool currently costs no credits; turning those lyrics into a song is charged separately. Check the price before each request. Failed generation tasks return deducted credits automatically; a completed song you dislike is not a failed task.
After supported vocal generation, ImgAI automatically requests lyric alignment. This follow-up can take longer or fail without affecting the audio. When timing is available, the LRC tool exports line timestamps; if it is missing, review the price before requesting alignment again.
WAV conversion, stem separation and MIDI transcription are separate actions with their own displayed prices. Stem results depend on the source mix. MIDI downloads are available only when the service returns a file; transcription can also return note data without a downloadable file. Review exports before using them in a DAW.