AI-coustics: Revolutionizing Audio Clarity with Generative AI Technology

With the numerous and increasing applications of AI in our everyday lives, it only makes sense to have it disrupt the audio industry. Introducing a unique technical approach that uses generative AI, AI-coustics brings a new tool to the table to tackle noise in audio recordings and improve quality.

Ever needed to listen to a recorded lecture or something and had to struggle with listening to the voices in that recording? That's practically what we sometimes experience with poorly recorded audio. And in content creation, obtaining the cleanest audio isn't always easy to do, and it can even be a bane for professionals with really noisy recordings.

Having automated audio processing tools does a ton to make the job easier. And in 2024, there is already a lot of them available. The experience with many of them, however, is that they aren't near perfect, and the type of audio you get when you put your recording through usually isn't pleasant. And with the AI powered ones, there is usually something off.

For example, when I use Adobe Podcast Ai to enhance my speech, I don't always like the outcome. It's clean and sounds professional, but it takes away the authenticity of voices sometimes. I didn't sound like me; it was more like the AI trying to sound like me.

Another thing about some of these "quick fix" audio tools is that their results aren't usually appealing. Sometimes, it mostly sounds muffled, and there may be obvious traces of the noise in the processed audio.

AI-coustics has a more nuanced technical approach to processing audio with their generative AI that do actual noise reduction work. “We developed a unique approach to simulate audio artifacts and problems — e.g. noise, reverberation, compression, band-limited microphones, distortion, clipping and so on — during the training process,” Fabian Seipel, co-founder and CEO AI-coustics, said.

Unlike other noise reduction methods that focus only on reducing background noise, AI-coustics addresses a wide range of audio artifacts and problems, which allows it to become more adept at handling a diverse array of audio issues. That results in enhanced clarity of voice.

AI-coustics uses a model trained on speech samples recorded in the startup’s studio in Berlin, AI-coustics’ home city. People are paid to record samples—Seipel wouldn’t say how much—that then get added to a data set to train AI-coustics’ noise-reducing model. _TechCrunch

It may feel as though people may begin to lose their jobs again in the audio industry. Pundits in the industry are concerned about how this new AI tool will affect them and their niche. Really, the tool, like many other AI tools, can be an augment for audio experts to make their job a lot smoother. In areas in their audio production process where deep and complex work isn't needed, AI-coustics can come in handy to make things easier.

“A content creation studio or broadcast manager can save time and money by automating parts of the audio production process with AI-coustics while maintaining the highest speech quality,” Seipel said. “Speech quality and intelligibility still is an annoying problem in nearly every consumer or pro-device as well as in content production or consumption. Every application where speech is being recorded, processed, or transmitted can potentially benefit from our technology.” He continued.

The model has been tested on audio from different scenarios: historical, lecture, interview, car drive, broadcasting, TV/movie, and aviation. All of which have different environmental contributions to the audio. And in the results, there is obvious clarity and improvement in the quality, while maintaining authenticity.

After using the model for a few tests, I am impressed with how well it performed despite the noisy environment that I was in. I can imagine how effective it will be with recordings in more quiet places.

If you are interested in giving the model a try, you can visit AI-coustics.com. You only get "sixty minutes of awesomeness" on a free account. For content creators that desire better audio, this is a game changer.