ElevenLabs releases Voice Isolator, an AI tool that removes background noise from voices for free

ElevenLabs developed AI voice-reading tools that reproduce the voices of famous people has an AI tool called ‘ Language Isolator ‘, which cleanly removes background noise from the voice. Voice Isolator can be tried for free on the demo site, so I was able to test for myself how effectively it can actually remove noise.

Free voice isolator and background noise remover | ElevenLabs
https://elevenlabs.io/voice-isolator

ElevenLabs launches free AI voice isolator to take on Adobe | VentureBeat
https://venturebeat.com/ai/elevenlabs-launches-free-ai-voice-isolator-to-take-on-adobe/

ElevenLabs has released a video showing what happens when you use Voice Isolator to remove noise from audio.

We just launched our Voice Isolator! 🚀

The best way to remove background noise and extract crystal clear dialogue from your content

Our madness of yesterday had method pic.twitter.com/FBExEXHIsn https://t.co/AtMuwUb4AL

— Ammaar Reshi (@ammaar) July 3, 2024

A man is speaking into a microphone while a fan is blowing strong wind towards him. The sound of the wind and the noise of the fan motor are constantly audible, making the sound very difficult to understand.

Use Voice Isolator to switch to a noise-free version of your audio.

The noise almost completely disappeared and only the man’s voice could be clearly heard.

On the ElevenLabs Demo page you can try “Voice Isolator”. The sound to be processed can be recorded directly with the device’s microphone or uploaded as a file. This time, click “UPLOAD AUDIO” to upload a file.

Select the audio file you want to remove noise from and click Open.

This is what the sound looks like before processing. You can hear sounds like footsteps and the sound of doors opening and closing.

I uploaded a 36 second audio file (312 KB) and it was processed in just a few seconds. If you want to listen to the processed audio file, click the play button. If you want to download the audio file, just click the download button.

The edited audio looks like this. The footsteps and door opening/closing sound that were present in the original file are gone, and only the speaking voice can be heard clearly. The audio you want to hear is preserved while only the noise is removed, making this a very accurate AI tool.

At the time of writing, Voice Isolator is only available through the ElevenLabs platform, but it is expected to become available through the API in the next few weeks. Voice Isolator can be used without registering or logging in as an ElevenLabs member, but processing costs 1,000 characters per minute of audio and is not available if you exceed the 10,000 character per month limit offered in ElevenLabs’ free plan. If you want to process longer audio files, you will need to sign up for a paid plan that costs $5 (about 800 yen) or more per month.

Technology media VentureBeat has noted that Voice Isolator is capable of removing random noise, which sets it apart from many tools that only remove certain types of noise. However, it is concerning that ElevenLabs has not released detailed information about the model underlying the AI ​​tool or whether the data fed into the tool is used in any way for training.