About a year ago, David Poblador began experimenting with synthetic voice generation. He had been working with audio and streaming technology for some time—he had spent 13 years at Spotify—and was starting to see that new artificial intelligence models could open up new possibilities. At the same time, Miquel Curanta was well acquainted with the other side of the coin: after years linked to the management of cultural companies and media, he knew how a newsroom works and, above all, which tasks are the most repetitive.
The point of convergence between the two gave shape to Rigora Labs and its first product, Rigora Bulletin. "We found ourselves at the right time," explains Poblador. The idea was to leverage the new generation of AI technologies to enter a very specific field, and one very familiar to Curanta: that of radio news bulletins and media audio content.
The proposal also has another differentiating element. "AI evolves very quickly and thousands of tools come out every day, but there are hardly any quality ones in minority languages," Curanta argues. Catalan, Basque, and Galician are their focus for a technology that aims to go further with other languages in similar situations. Catalan is the first step, and "AI in Catalan usually speaks like a robot; with us, it speaks like a person," Curanta summarizes.
Natural voice generation is, therefore, an essential part of the product. However, it is not the only one; Rigora Bulletin is designed to automate the process of creating an audio bulletin from content that a media outlet has already produced. The platform can convert published news into a script adapted to radio language and generate the audio, with the possibility of integrating different sources and distribution systems.
Although their technology has a direct impact on how newsrooms work, Curanta emphasizes that "the project does not come to replace the work of journalists, but to complement it." The difference, according to the founders, is what remains in the hands of the newsroom: Rigora Bulletin does not create news or do street journalism: it starts from the information that the media has already prepared and transforms it into an audio product. The journalist continues to decide what information is published and can intervene in the result, but the technical preparation and narration can be automated.
"We come to complement the work with an application that is useful for a media outlet," adds Curanta: "Journalists have to be on the street, making news. They don't have to be in the studio waiting for the 12 o'clock bulletin. We make bulletins based on journalism; we don't write content, but rather we use what a journalist has done, and we convert it into audio content."
For Poblador, the key has been to understand that it was not simply about applying an artificial voice to a text. "The crucial part is the product," he explains. A bulletin has a ladder, sections, and its own routines. Meteorology, for example, can be automated from certain sources and incorporated into the bulletin with a defined radio style. The same principle could be applied in the future to sections such as traffic or other news blocks.
Now, the technology is already functional, and the founders are testing it with several media outlets. One of the current challenges is to integrate Rigora Bulletin into the workflows already used by broadcasters: "If your radio receives the files that need to be broadcast in MP3 and they are saved in Dropbox, we want to be integrated into the different ways of working," explains Poblador.
The product is in the testing phase and the founders plan to start commercialization soon. The pricing model will include different levels according to the volume of use, with an approximate range of between 50 and 500 euros per month. With this proposal, it is aimed at institutions and mainly at media outlets: in Catalonia, there are a thousand; in Spain, 4,000. But the universe on which it focuses is much broader, starting with Europe, and looking at the whole world: "We want to be a universally accessible application."
With this ambition, Catalan is the starting point, and it has already generated bulletins in Basque and Spanish. It is also conducting tests in other languages, such as Swedish and Dutch, with the conviction that solving the difficulties of a language with fewer technological resources can also become a gateway to other markets.
Thus, the project that began with two profiles coming from different worlds seeks to convert AI into another piece of media infrastructure; not to decide what is news, but to automate a part based on the work done by the newsroom.