Several days ago, the AI company ElevenLabs launched its latest tool called Voice Isolator which lets you remove background noise, leaving you with clear dialogue. Now, the company has launched the Voice Isolator API, which lets third-party developers integrate Voice Isolator into their applications to come up with new applications and services.
The company only put out a very short statement to announce the API, it said that the API is charged at 1,000 characters per minute of audio and that technical information can be found in the documentation. It published a demonstration video with the statement too.
The demo shows the creation of a website in Claude which can pull in the audio from a user-provided YouTube video and then have the audio cleaned up by Voice Isolator. The resulting website looks pretty good and is very straightforward for the user to interact with, there is a box for the YouTube link, a button to generate the cleaned audio, and then a button to play the audio back. There is then an audio progress bar to see how much of the clip is left.
Obviously this use case is just a demo but it will be up to third-party developers to find creative ways to use Voice Isolator in their own products. Inadvertently, the demonstration also shows how good AI bots like Claude are becoming at programming tasks, in just one detailed query, it was able to generate the code for the web page.
The Claude model that was used in the demo was the latest Claude 3.5 Sonnet model which surpassed both GPT-4o and Gemini 1.5 Pro in most benchmarks. Sonnet is the middle-powered model that balances results with speed. Opus and Haiku are the bigger and larger models respectively but they"ve not been released for Claude 3.5 yet.