ElevenLabs used Shutterstock’s diverse sound effects to power its AI sound effect generator. Discover how they created versatile soundscapes at scale.
In this post, we’ll cover:
ElevenLabs’ Groundbreaking AI Sound Effect Generator
Where does a company shaping the future of on-demand sound turn to when it needs to create the next disruptive AI audio technology? ElevenLabs tapped into Shutterstock’s high-quality, large-scale datasets to build its state-of-the-art AI sound effect generator.
ElevenLabs began with AI tools for creating human-like voices and expanded into audio AI with its sound effects model. The new tool uses text prompts, letting creators describe sounds instead of searching through libraries. This makes it faster and more precise for creators in film, TV, games, and social media to build soundscapes.
Perfecting AI-Generated Sound Effects
To build the sound effect generator, ElevenLabs needed to fine-tune its AI. Partnering with Shutterstock gave access to the right content to train the model for accuracy.
Audio content—when paired with rich data—plays a significant role in generative music and sound effects, particularly in areas like speech recognition, audio generation, and sound analysis. It provides AI tools with the necessary information to recognize, generate, and enhance audio interactions, making applications in speech, music, and sound much more effective.
After fine-tuning their model on Shutterstock’s licensed audio, ElevenLabs built a tool that allows creators to generate sound effects directly from text prompts. Instead of searching for pre-made tracks, users can describe what they need, and the model produces tailored options.
This speeds up the creative process and offers more control over the final project.
Customized Care and Service
Shutterstock’s Data Partnership team went the extra mile to ensure ElevenLabs’ success, engaging in in-depth discussions to gain a comprehensive understanding of its vision and goals. By exploring the intricacies of its AI audio generation project, Shutterstock was able to tailor its data offerings to align perfectly with ElevenLabs’ unique needs.
The process goes fairly quickly, too. In ElevenLabs’ case, it took approximately two weeks to sign the contract and receive the data. The team appreciated that Shutterstock has a long history of working with creative partners and has a system which compensates contributors for uses of their content through its Contributor Fund.
“Shutterstock helped us source a wide variety of high-quality sound effects that was key to improving our sound effects model,” says Alex George, Head of Data at ElevenLabs. “Now, creatives around the world can generate any sound they need to tell their stories, with more control and flexibility in their projects.”
Another benefit of working with Shutterstock is its experience transferring data via an Amazon Web Services’ Simple Storage Service (S3) bucket. The object storage service is relied on because it offers industry-leading scalability, data availability, security, and performance.
If you want to learn more about its data services, you can contact Shutterstock directly or work through Amazon Web Services and Google Cloud. The team is happy to help you ideate, curate, and customize datasets for your needs.
A Partnership Turned up to 11
Now that the text-to-sound effects model is available, it’s an entertaining playground for the ears. Users just describe the sound they have in mind and it will generate some samples to choose from, which can then upscale to the one they prefer most. ElevenLabs is eager to see more games, films, and audio stories enhanced with generative sound effects.
This collaboration underscores the power of rich, well-structured data in pushing the boundaries of AI technology. By integrating generative music and sound effects into its suite of products, ElevenLabs continues to support creators across industries in crafting immersive, one-of-a-kind audio experiences.
License this cover image via Jade ThaiCatwalk.
This post was originally published onNovember 4, 2024
Recently viewed
${excerpt}