The Campaign

With a custom machine learning algorithm that analyses and completely recreates a person’s voice, Project Revoice gives ALS patients the full use of their own authentic voices, allowing them to speak freely and naturally even after they physically can’t. To launch the program we recreated the voice of Pat Quinn, who co-founded the ALS Ice Bucket Challenge while fighting the disease himself.

Creative Execution

In March 2018, surrounded by his family and friends Pat was finally able to speak again in his own voice, after over a year in silence. This emotional moment became the heart of our launch campaign, which started in our owned channels only but quickly spread through social conversations and earned media, primarily in US, Canada, UK and Australian markets. To make this technology available to the wider ALS community, Project Revoice has built an online voice bank where ALS sufferers can easily record the voice material necessary for their own voice clones. By the end of 2018 these recordings will be used to create more ‘Revoices’, which users can access and ‘speak’ with through the website or a custom API, via their own Assisted/Augmentative Communication (AAC) devices.

Since launch, Project Revoice has received significant interest from both the ALS community and the general public: Over 900 million earned media reach Over one million organic video views in the first week. By week two over 41 million people had posted, commented shared or mentioned the campaign. Over 680 articles written globally $9.8 million estimated earned PR value. Over 500 patients joined the program in the first month alone.

To rebuild Pat’s voice we had to create a data bank of audio for the algorithm to work with. Since Pat hadn’t banked any audio himself, our challenge was to build a sufficient data bank from old Ice Bucket interviews and speeches found online. In a sense, it was only because Pat gave the ALS community a voice through the Ice Bucket Challenge that we were able to give him his own voice back. After sourcing over 100 individual files of varying quality, we manually analyzed, cleaned up and transcribed every usable section to create a unified data bank for the algorithm to work with. Once the voice had been cloned through Lyrebird technology, we built a custom user interface for Pat to access the voice through his eye-tracking computer.