Stable Audio: The Future of Text-to-Audio AI by Stability AI

In a groundbreaking move, London-based generative AI pioneer, Stability AI has launched its innovative text-to-audio AI platform, “Stable Audio.” Stability AI makes an exciting foray into the world of music and sound generation with this cutting-edge platform that uses the most recent in artificial intelligence. With the capability to produce gripping songs up to 90 seconds, it’s an ideal solution for diverse projects, from commercials and audiobooks to video games.

While Stability AI has always been at the forefront of AI-driven visuals, the introduction of “Stable Audio” places them in direct competition with tech giants like OpenAI, Google, and Meta. The Stable Audio platform leverages a diffusion model, similar to their renowned image platform, Stable Diffusion. However, the twist is in its training—it’s been extensively trained with audio data, not images. This shift allows users to instantly generate songs or background audio of any length, making it a versatile tool for various projects.

Addressing the flaws of traditional audio diffusion models, Stable Audio has undergone extensive music-specific training. It incorporates detailed text metadata, specifying the start and end times of a song. This feature offers users unmatched flexibility in song length, granting them complete control over their creative endeavors.

Stable Audio is a testament to the innovative audio generation research at Stability AI’s Harmonies Lab,” the company announced. “We’re constantly refining our model architectures, datasets, and training methods to enhance output quality, controllability, speed, and length.”

The platform boasts training on an expansive dataset of over 800,000 audio files, spanning music, sound effects, and individual instrument stems. This dataset, which includes text metadata from AudioSparx, covers an impressive 19,500 hours of diverse sounds. Stability AI has also ensured all legalities by securing permissions for copyrighted materials through partnerships.

For users eager to explore, Stability Audio offers three pricing tiers:

1. Free Version: Generate up to 45 seconds of audio, with a limit of 20 tracks per month.
2. Professional Tier: At $11.99, users can create 500 tracks, each lasting up to 90 seconds.
3. Enterprise Subscription: Customized plans and pricing structures for businesses

It’s worth noting that the free version restricts commercial use of the generated audio.

While text-to-audio generation isn’t novel, Stability AI’s approach is revolutionary. Other tech giants, like Meta with AudioCraft and Google with MusicLM, have dabbled in this space. However, their offerings are currently limited to select professionals and researchers.

from EkTechNerd.Com https://ektechnerd.com/stable-audio-the-future-of-text-to-audio-ai-by-stability-ai/
via IFTTT