Try Stable Audio v2.0

Summary

• On April 3, Stable Audio 2.0, which automatically generates high-quality music and audio, is now available from Stability AI.
  • 44.1 kHz stereo music can be generated
  • Maximum 3 minutes per song
  • Text to Audio and Audio to Audio also available.
  • 20 credits can be used for free, and each time a song is generated, 1-2 credits are consumed.
• API is said to be released in the near future.

Try it out(Text To Audio)

• Input Data
  • Prompt
  • Prompt Library
    • Select a song genre and the basic spell is automatically set in the Prompt.
  • Model
    • Stable Audio Audiosparx 1.0：1 credit consumption
    • Stable Audio Audiosparx 2.0：High quality. 2 credit consumption. If you want to make a song of maximum 3 minutes, this is the only way to do it.
  • Duration
    • The length of the song can be specified.
  • Input audio
    • Can upload homegrown songs and tunes. Looks like you can make a song based on this.
  • Add extras
    • Steps
      • Maybe a higher number of Steps will produce a higher quality song at the expense of generation time.
    • Seed
      • When Seed is fixed, reproducibility appears in the generated songs.
    • Prompt strength
      • Maybe you can set the intensity of how much to comply with the spell.

Impressions

• Vocal generation is not supported at this time.
• Compared to Suno, Stable Audio still seems less practical.
• The feeling of inputting a prompt is similar to Stable Diffusion, an image generation AI.
• I feel that the Prompt regarding the music is very detailed, so I think I will have a hard time until I get the desired music.