- AI company Resemble AI has developed Resemble Enhance, an open-source tool that improves the quality of historical audio.
- The software uses a two-step process – first, a sophisticated denoiser removes static background sounds, then an AI-driven speech enhancer enriches the audio.
- Resemble Enhance is a free of charge solution that can be particularly beneficial for podcasters, entertainment industries and educational settings, as well as reviving historical recordings.
In the context of many artificial intelligence firms focusing on technology to develop or improve video, there is one company that’s emphasizing audio fidelity: Resemble AI. They’ve created an open-source tool, Resemble Enhance, designed to upgrade the quality of historical audio dramatically. The tool can take a distorted, low-quality historical speech recording and make it sound as if it were more recently recorded or broadcast.
Resemble AI distinguishes Resemble Enhance through its dual-module approach, consisting of a sophisticated denoiser and an AI-powered speech enhancer. The denoiser removes static background noises, while the speech enhancer boosts the overall audio quality.
The Canadian company’s denoiser uses UNet, an AI model, to separate the different types of sounds in a recording. Once the UNet has filtered out unwanted noise, leaving only the speech, the enhancer module is then activated to extend audio bandwidth and correct distortions. The company claims that this dual functionality ensures the final audio output to be not only noise-free but also possesses the richness of contemporary recordings.
One of the striking features of Resemble Enhance is it being open-source, which means it is accessible for free. It notably offers an affordable way to restore media in a traditionally high-cost market of media restoration services. The primary users of Resemble Enhance are likely to be industries that rely heavily on clear audio quality, including podcasting, entertainment, and education.
The demand for high-quality digital content is rising; hence, this open-source nature could make Resemble Enhance a potentially disruptive tool currently dominated by expensive proprietary solutions. With the merging of AI in audio and video enhancement, comprehensive media restoration solutions may likely be paving their way.
For users wanting to experiment with Resemble Enhance, they can visit the official Resemble AI website or download the models from the project’s official Github page.