Dubai startup helps AI-voice cloning take centre stage

Wam

• Camb.ai tool supports 120 languages
• Creators see end of language barrier
• New markets for sports and entertainment

A YouTube video released in January of Novak Djokovic shows the tennis star speaking fluent Spanish at a post-match press conference. 

While the Serbian champion is renowned for speaking several languages, in this case his voice was created by an artificial intelligence tool developed by a Dubai-based startup, Camb.ai.

The original Djokovic footage was taken at the Australian Open tennis tournament. Organisers also used the translation software developed by Camb.ai to create simulated lip movements for other players speaking a range of languages, such as Coco Gauff and Daniil Medvedev.

---

---

Showcased during the tournament’s “demo day”, the results garnered mixed reactions from commentators, with one saying it was “creeping me out”.

Traditionally, companies relied on tools such as Google Translate to handle speech-to-text conversion followed by text-to-speech conversion to allow people to communicate in languages other than their native tongue.

“This approach often fell short in capturing the intricacies of human speech, resulting in robotic outputs similar to outdated technology,” said Camb.ai co-founder Akshat Prakash.

• Making sense of Arabic machine translation
• Saudi Arabia steps up battle against AI cyberattacks
• OpenAI in talks with Abu Dhabi’s G42 for up to $10bn funding

Camb.ai CEO Avneesh Prakash highlights the difficulties faced by content creators in reaching broader audiences due to translation barriers.

“Content creators, especially Arabs, are increasing and getting more aspirational and want to get out of the region – they want to go global. However, language is a barrier since many do not understand Arabic,” he said.

In response, new advancements in AI technology such as video translation tools have emerged. Creators can now dub, lip sync and even clone their voices in various languages, opening new horizons for content localisation and global audiences.

Camb.ai has collaborated with Nayla Al Khaja, the UAE’s first female Emirati film director, to release her movie Three in multiple languages.

The AI “generates dialogues in over 120 languages and accents, using the voices of the artists,” said Al Khaja.

“The significance of using AI in an independent film to amplify cinematic narrative cannot be overstated, as it becomes the catalyst for unlocking new horizons for films and entertainment,” she told AGBI.

The technology is currently being used to show Al Khaja’s film in Mandarin in theatres in the UAE.

Democratising language

Funded by the Dubai Future District Fund, Camb.ai offers foundational models in text-to-speech and language translation through voice cloning. In the case of the Australian Open it delivered the outcome in under 12 hours during the tournament.

Its platform supports over 120 languages including many Arabic dialects. 

“We intend to democratise language access, whether recorded, archived, or streamed, making alternate language selection a matter of choice rather than privilege,” says Akshat.

With over 7,000 languages spoken globally, each with its own dialects, bridging linguistic gaps is a daunting task.

In regions like the Middle East, Arabic may be the dominant language but there are dozens of dialects and millions of unique words so the challenge is higher. 

However, the integration of AI in language translation has sparked ethical and privacy issues, raising questions about concerns and data privacy.

“Organisations must prioritise ownership, privacy and data protection, ensuring that individuals rights are respected,” adds Avneesh.