Songs that contain profanity or explicit content are often censored for public listening over the radio. Some artists create alternative versions with different words, known as "clean edits". However, for the majority of songs, clean edits do not exist, leaving radio stations to censor them manually by skipping the beat or adding a bleep. Eco Clean aims to utilize AI digital technology to provide radio stations with an AI-generated voice library of alternative words to replace bad ones and clean any song. This digital Audio Cleaning L[AI]brary allowed stations to play any song smoothly without intrusive bleeps.

Many songs contain profanity, which is not appropriate for all audiences. To address this, some artists create "clean edits" of their songs with alternative lyrics that are suitable for public consumption. However, for most songs, clean edits do not exist, which means that radio stations have to manually censor them by skipping over the explicit parts or using a bleep sound. This is done to ensure that the songs are safe for everyone to listen to. In Jordan, as in other parts of the region, it is common for older relatives or teachers to threaten to wash the mouths of children who use inappropriate language. This cultural insight inspired a local fabric cleaning brand to create a campaign where they "wash" the mouths of artists, while still preserving the original flow of their songs.

Eco Clean from GIG is a local brand offering laundry detergent, fabric softener, and stain remover committed to cleaning all of life’s messes, including the most stubborn or invisible ones. From one Jordanian household to another, we understand the challenges of going outside and staying spotless. How can we show our commitment to keeping our ecosystem spotless?

Can we demonstrate our cleaning powers in a medium that is not visual? We found our opportunity in the songs played on the radio, and how radio stations use disruptive bleeps to cover explicit content instead of cleaning it.

So, we pitched in to help with that.

First, we gathered lyrics data from popular songs to identify the most commonly used R-rated words and phrases. Then, we used AI technology to generate a list of alternative rhyming words that have the same number of syllables as the original words or phrases. This allowed us to create a library of voice bites that can be easily inserted into any audio clip, seamlessly replacing male or female voices in any genre.

We purchased previously unbuyable bleep time and shared the Audio Cleaning L[AI]brary files with radio stations. Collaborating with the radio's audio engineer, we replaced the profane words and inserted the new words into the vocal track, ensuring a seamless fit by adjusting speed, pitch, and voice matching. This process was repeated several times to achieve the best outcome.

Before each 'cleaned' song, a brief intro crediting Eco Clean for cleaning the song was played.

We cleaned over 500 words in more than 120 songs spanning all genres over a 30-day period. With an average song duration of 5 minutes, we delivered over 18,000 minutes of spotless tunes to various radio stations.