ElevenLabs, the AI voice startup known for its voice cloning, text-to-speech, and speech-to-speech models, has just launched a new tool: an AI Voice Isolator. Now available on the ElevenLabs platform, this tool allows creators to remove unwanted ambient noise and sounds from any content, including films, podcasts, and YouTube videos.
A New Tool in the Creative Arsenal
The AI Voice Isolator arrives shortly after ElevenLabs released its Reader app. While the tool is free to use with some limitations, it’s worth noting that enhancing speech quality is not a novel capability. Many other providers, including Adobe, offer similar tools. However, the true test will be how well Voice Isolator performs compared to these existing solutions.
How Does the AI Voice Isolator Work?
Creators often face the challenge of background noise when recording content like films, podcasts, or interviews. These noises can interfere with the final output, diminishing the quality of the recorded speech. Traditional solutions, such as using microphones with ambient noise cancellation, can be costly and out of reach for early-stage creators with limited resources.
This is where the AI Voice Isolator steps in. During the post-production stage, users upload the content they want to enhance. The tool then processes the file, detects and removes unwanted noise, and extracts clear dialogue. ElevenLabs claims the product can deliver speech quality comparable to studio recordings. In a demo, the company’s head of design, Ammaar Reshi, showcased how the tool effectively removed the noise of a leaf blower, leaving crystal-clear speech.
Real-World Testing
We conducted three tests to evaluate the Voice Isolator’s real-world applicability. In the first test, we spoke three separate sentences with different background noises. The other two tests involved sentences with a mix of various noises occurring randomly.
In every case, the tool processed the audio within seconds. It successfully removed noises like door openings and closings, table banging, clapping, and household item movements, extracting clear speech without distortion. However, it struggled with wall banging and finger snapping sounds.
Limitations and Future Improvements
Sam Sklar, who handles growth at ElevenLabs, noted that the tool does not currently work on music vocals, although users are encouraged to experiment with it.
While the AI Voice Isolator’s ability to remove irregular background noise is impressive, there is still room for improvement. Ongoing enhancements are expected, as with other tools. However, details about the underlying models powering the tool and whether recordings are used for training remain undisclosed. Users can opt-out of data use for training via a form linked in the company’s privacy policy.
Access and Pricing
Currently, the Voice Isolator is available only through the ElevenLabs platform, with plans to open API access in the coming weeks. Free access is available with certain usage limits—10k characters per month, translating to approximately 10 minutes of audio per month. For larger audio files, paid plans start at $5/month.
