Speaking my language: Bringing AI to the fore with real time audio translation for French football

Ligue de Football Professional’s commercial entity is committing to provide real time multilingual accessibility to French football fans in more than 160 languages, using AI from Camb.AI. Pictured is Akshat Prakash, co-founder and CTO of Camb.AI

February’s SVG Europe Scaling up for DTC online event issued a challenge for the sports broadcast industry to push the boundaries of audio personalisation.

“As leaders in the audio industry, I think we should be pushing things further,” said Salsa Sound’s CEO Rob Oldfield. “Why can’t we leverage software tools and aim a bit higher? DTC is a great test bed for that and one of the beauties for me about DTC is that you can explore some of these more personalised experiences and give a bit of control back to the consumer.”

It was a bold challenge, and it is exactly what LFP Media is aiming to do for millions of French football fans across the globe. Announcing a multi-year deal with speech and translation specialist Camb.AI, Ligue de Football Professional’s commercial entity is committing to provide real time multilingual accessibility to French football fans in more than 160 languages, using artificial intelligence (AI).

As the governing body overseeing Ligue 1 McDonald’s and Ligue 2 BKT, LFP Media represents 36 professional clubs across France. Ligue 1 McDonald’s is the most followed sports league in the country and has a global following of over 50 million fans on social media. By integrating Camb.AI’s AI translation, voice emulation, and dubbing technologies, the partnership enables Ligue 1 McDonald’s global fanbase to enjoy match broadcasts, commentary, interviews, and exclusive content in any language they choose.

Action replay

It is not the first time Camb.AI has worked directly with sports broadcasters to deliver linguistic inclusivity; the company has previously partnered with Major League Soccer, Tennis Australia and Maple Leaf Sports & Entertainment. In fact, its collaboration with Major League Soccer delivered the first-ever AI-translated live commentary in sports. More recently Camb.AI worked with Eurovision Sport to test live AI-generated translated commentary at the 2024 World Athletics U20 Championships in Lima.

“We continue to partner with Eurovision Sports, for whom we’ve done events like the World Athletics and World Indoor Championships, and we started working on the LFP Media partnership with several tests and POCs to evaluate the output quality for several localisation use cases,” says Akshat Prakash, co-founder and CTO of Camb.AI.

“Camb.AI offers more than 160 languages, including underserved indigenous ones backed by more than six years of research in generative and speech AI. Our MARS6 speech model was also the first generative voice model on AWS Bedrock, reflecting the pioneering work we’re doing.

“Today, Camb.AI livestreams, on average, six to eight hours of live content a day, in multiple languages for sports leagues, brands, and media houses. The value that Camb.AI brings to the entire workflow of sports broadcasting, to deliver games in multiple languages, is unmatched.”

Setting a precedent

It is one of those applications of AI where it is easy to see the benefit and Camb.AI says LFP Media’s vision of making French football more inclusive and accessible aims to set a precedent across European sports. For LFP Media, it is all part of a long term gameplan.

“We have worked with Camb.AI over the last few months to identify multiple use cases and impact points which enhance our fan engagement and will bring unprecedented value to our international broadcasters,” says LFP Media’s chief media officer, Martin Aurenche. “This partnership marks another step in LFP Media’s strategy to become the world’s leading league in AI-driven innovation.”

Built by an Interspeech-published team from Carnegie Mellon, Apple, and Amazon, Camb.AI ’s MARS6 is a frontier text-to-speech (TTS) model included as the first voice model on Amazon Web Services’ (AWS) Bedrock platform.

Prakash says that when Camb-AI introduced MARS6 on AWS Bedrock, it gained its first model that was capable of speech-voice synthesis, introducing a whole new modality in generative AI to AWS Bedrock. It’s significant as it means that AWS Enterprises and developers can use MARS6 in their AWS applications and deployments. Meanwhile, the combination of the company’s Boli – translation – and Mars – speech emulation – models allow content to be translated into other languages without losing the speaker’s tone, emotion, and passion.

Seamless integration

Moreover, Prakash says its implementation is agnostic to a broadcaster’s infrastructure. “We can simply ingest feeds (SRT, HLS, RTMP, or otherwise) and send out the same feed in multiple languages,” he says. “There is no perceived latency for viewers listening in alternate languages and the video and alternate audio track are always synced. This simple, stateless integration makes it possible for Camb.AI’s technology to embed in the most complex of streaming infrastructures as well, regardless of the downstream complexity.”

Technology like this feels like a significant step on the road to audio personalisation, and the conclusions drawn at February’s SVG Europe Audio Summit feel on-the-money, but like everything in sports, it all comes down to teamwork.

“The development of generative AI is never behind closed doors; it requires the partnership of risk-takers and people willing to believe in a dream,” confirms Prakash. “We’ve been lucky to have partnered with a lot of great companies, enterprises and sports leagues that have helped us iterate on our technology and start making it truly production-ready.

“We think this is one of the most impactful applications of AI and it has the potential to bring LFP and other partners to every last culture in the world. It’s exciting to see that sports are now actually on their way to becoming universal.”

 

Subscribe and Get SVG Europe Newsletters