SongGeneration generates music via the command line or via scripts. The core operation is to prepare an input file and run an inference script. The input file should be in JSON Lines (.jsonl) format, where each line represents a generation request and contains the following fields:
- idx: The filename (unique identifier) of the generated audio.
- gt_lyric: lyrics in the format
[结构] 歌词文本e.g.[Verse] 这是第一段歌词The - descriptions (optional): Describe musical attributes such as
female, pop, sad, piano, the bpm is 125The - prompt_audio_path (optional): 10-second reference audio path for imitating styles.
After running the inference script, the model generates the complete song based on the input.
This answer comes from the articleSongGeneration: open-source AI model for generating high-quality music and lyricsThe
May not be reproduced without permission:AI productivity tools " How does SongGeneration generate songs? What inputs are required?
English 
简体中文 
日本語 
Deutsch 
Português do Brasil