DiffRhythm AI

DiffRhythm AI is a web-based song generation tool built on DiffRhythm, described on the page as the first latent diffusion-based song generation model capable of synthesizing complete songs with both vocals and accompaniment for up to 4 minutes. Users enter lyrics and a style prompt, then generate a song that can be downloaded and shared. The page states the model uses a non-autoregressive structure for fast inference, and that it originates from research by the ASLP Lab, though DiffRhythm.ai is not affiliated with ASLP Lab. The site offers V1 through V4 models, Simple and Custom modes, an Instrumental Mode option, and a gallery of AI-generated tracks. Subscription plans are listed with credit-based usage, and a free plan with a one-time quota of 8 songs is offered.

Key features

• Generates complete songs with vocals and accompaniment from lyrics and a style prompt
• Style control through simple text prompts across genres
• Non-autoregressive structure for fast generation of songs up to 4 minutes
• Simple and Custom generation modes
• Instrumental Mode option
• Download in WAV and MP3 formats on paid plans
• Extract stems and remove vocals on listed plans
• Commercial license on listed plans
• Gallery of latest AI-generated music examples

Best suited for

• Musicians and creators generating song drafts from lyrics
• Businesses seeking music for commercial use on applicable plans
• Users wanting quick full-length song prototypes with a style prompt

Pricing

Free plan with a one-time quota of 8 songs ($0/month). Starter plan listed at $6.99/month (billed monthly, 18,000 credits) or $83.9/year billed yearly. Unlimited plan listed at $59/month or $712.9/year billed yearly. Monthly and yearly billing discounts are shown on the page.