Song and Sensibility

With the recent release of the second single, 'Chinna Chinna Kangal' from Vijay's The Greatest of All Time, film audiences have been treated to a double delight with the song featuring both the voices of Vijay and late singer Bhavatharini. Composer Yuvan Shankar Raja and director Venkat Prabhu employed innovative AI technology from Pitch Innovations, the company that also brought back the voices of Bamba Bakiya and Shahul Hameed for the 'Thimiri Yezhuda' song in Lal Salaam. Notably, TR Krishna Chetan, CEO of Pitch Innovations, reveals they created an AI voice model of Bamba Bakiya even before AI became mainstream.

For 'Chinna Chinna Kangal,' Krishna says recreating Bhavatharini's voice required specific recordings. "As we are doing a studio-quality AI model, we need original tracks of the singer's voice recording, without any sound editing such as reverb or EQ," he explains regarding the first step for the model. "We need the singer's voice in different pitches, and multiple styles, such as a happy or sad song," he adds.

Krishna further elaborates on the process: "The model also requires a singer to sing a track version of the song, for it to be replicated in the desired singer's voice." He reveals, "Priyanka NK sang the track version," adding that she was informed beforehand that her voice would be replaced with the AI voice of Bhavatharini.

Composers have often received flak for the ethical implications and problems that AI technology comes with. On the conversation around AI and ethics, Krishna says, "Today, the AI models of top Hindi singers are readily available online, and it is being used by many illegally. We ensure to do only projects that follow ethics. We ensure that the composer or producer has rights to the artists' voice and that the singer who is being used is paid for the same."

One main ethical problem that is often cited is that anybody, even record labels and film studios, with access to the technology can use the voice without the consent of the demised. But Krishna reveals that most artists he has interacted with are happy about the thought that their art can continue even after their time. "I knew Bamba Bakiya sir personally before his demise and I can surely say that he would have been happy about his voice being used even after his time." He then adds, "At the end of the day, we are only the service provider, and from our side, we are ensuring to follow the law and do it ethically."

On fears that AI might leave people out of a job, Krishna says that audio manipulation AI is a technology that is more of an assistive tool. He explains quoting an example," A composer auditions a few singers before finalising one singer for a song. Apart from the composer and the singer, the time and effort of sound engineers and other technicians are also wasted in this process. With AI, every singer can get an AI model of their voice, and with this, the composer can try different singers and finalise easily, without wasting resources." He further adds that this technology would create opportunities for debut singers, rather than foil them. "Even for new singers, this technology will help them audition at several music studios, even transcending national boundaries, and expanding their scope of opportunities," he concludes.