cross pond high tech

AI can generate storyboard animations from scripts, spot potholes and cracks in roads, and teach four-legged robots to recover when they fall. But what about adapting one person’s singing style to that of another? Yep — it’s got that down pat, too. In a paper published on the preprint server Arxiv.org (“Unsupervised Singing Voice Conversion“), scientists at Facebook AI Research and Tel Aviv University describe a system that directly converts audio of one singer to the voice of another. All the more impressive, it’s unsupervised, meaning it’s able to perform the conversion from unclassified, unannotated data it hasn’t previously encountered.

The team claims that their model was able to learn to convert between singers from just 5-30 minutes of their singing voices, thanks in part to an innovative training scheme and data augmentation technique.

“[Our approach] could lead, for example, to the ability to free oneself from some of the limitations of one’s own voice,” the paper’s authors wrote. “The proposed network is not conditioned on the text or on the notes [and doesn’t] require parallel training data between the various singers, nor [does it] employ a transcript of the audio to either text … or to musical notes … While existing pitch correction methods … correct local pitch shifts, our work offers flexibility along the other voice

While the Phantom is a product that’s actually for sale in places such as the Apple store, its real purpose is to showcase Devialet’s audio technologies. In a recent interview, Devialet CEO Quentin Sannié explained that the company’s ultimate goal is to get its technology into everything from phones, to cars, to television sets — really, anything that outputs audio.

The Phantom is a showcase for two of Devialet’s developments: a software-based technology called Speaker Active Matching (SAM) and a hardware-based amplification system that combines an analog amp with a digital one. Sannié says that the SAM algorithm could be implemented into any digital signal processor, even the one in your phone, and can improve the sound quality of hundreds of devices. The algorithm is developed by measuring the behavior of each speaker and then creating a mathematical model to control how the speaker behaves. This measurement process can be done in about 10 minutes using special, laser-based tools the company developed for it.

The Phantom is a showcase for Devialet's technologies

Sannié says that a system using SAM can reproduce lower frequencies without changing its hardware at all, and it can even enable noise-cancellation without the need for a subwoofer. It’s part of the reason that the Phantom’s relatively small drivers move so quickly when the volume is cranked up and have little to no distortion. Sannié claims that SAM could even be used by a streaming service to optimize audio before it’s even delivered to a listener. A user could tell the service what kind of speakers or headphones they are using and the sound would be tailored just for them.