Resemble AI is concentrated on audio fidelity, but many AI startups are racing to discover ways to use the technology to improve or even create video. The startup has released ‘Resemble Enhance,’ an open-source tool to greatly improve the quality of historical audio. It also provides businesses with an AI voice generator to generate realistic human-like voiceovers.

The new service can apply artificial intelligence (AI) to a distorted, fuzzy recording of a long-lost historical speech, making it sound as though it was recorded or broadcast yesterday.

Resemble Enhance, according to the Canadian company, stands out for its dual-module approach, which combines an AI-powered speech enhancer with a sophisticated denoiser that eliminates static background hums and hisses. This combination improves the audio’s overall quality while also eliminating unwanted noise.

Despite the fact that there are other audio restoration products available, Resemble’s mix of methods might make a significant difference.

How does it operate?

UNet, an AI model, is used by the Resemble AI denoiser to assist in differentiating between the various sound types that can be heard on a recording. It is very good at eliminating background noise from audio files so that the speech is the only thing that is heard.

The enhancer module starts working after UNet completes its task, expanding audio bandwidth and removing distortions. According to the company, this dual functionality guarantees that the output is both noise-free and has the richness of modern recordings.

Resemble Enhance is a free tool that is available as open-source software, making it a strong choice in the typically pricey media restoration services market. Industries like podcasting, entertainment, and education that depend on high-quality audio are the main winners from Resemble Enhance. Furthermore, this tool revitalizes historical recordings, which may lead to a better understanding of the past.

The tool is being released at a time when there is a record demand for superior digital content. Meanwhile, Resemble Enhance’s open-source design makes it a potentially disruptive force in a market where expensive proprietary solutions currently hold a monopoly.

More complete media restoration solutions will probably be made possible by AI’s convergence in audio and video enhancement.

Users can now obtain professional results with their own computers for a minimal investment by combining this tool with other video enhancers, such as GPEN or the well-known GFPGan, that use generative AI or other models to upscale and enhance images and faces.