We didn’t expect this to happen so soon, or maybe not at all.
Our founder, the late Satguru Sivaya Subramuniyaswami, always had the intention of recording himself speaking the edited and published version of his Master Course book trilogy lessons, which consisted of three 1,008-page books – Dancing with Siva, Living with Siva, and Merging with Siva. However, due to his busy life of service and sadhana, recording already-published material was not a top priority. In his final years, he recorded over a hundred lessons from Merging with Siva. He had the monks set up a high-end recording station, and every afternoon he would spend an hour or two reading the lessons. This went on for several months. Then, quite suddenly in 2001, after returning from a trip with Innersearchers through Northern Europe, he learned that it was time to release his physical body. Siva’s Will Be Done.
We produced an audio CD of a few chapters from those Merging with Siva recordings and contented ourselves with that and Gurudeva’s other spontaneous audio talks for several years. Occasionally, we tried to replicate his voice, but no one could quite get it right. So we waited.
Just last year, artificial intelligence (AI) capabilities advanced significantly. We came across articles about voice cloning AI systems, and considering the potential fulfillment of Gurudeva’s original intention, we delved into it. After trying out various AIs, we found the ElevenLabs model to be quite promising. While it wasn’t perfect initially, their model has since evolved rapidly, and the results are now quite impressive. We have successfully replicated Gurudeva’s voice to the point where those who hear it and knew him are amazed at how accurate it sounds.
This achievement was made possible because we had high-quality recordings of Gurudeva speaking over 100 Merging with Siva lessons. By feeding a couple of hours of those recordings into the AI, we now have a voice clone that can speak any text we input. Though technically artificial, we believe the spiritual energy is still present as it mimics the exact tones and style of his original voice.
Our plan is to gradually create audio files of all the Master Course trilogy lessons, integrating them into web pages and daily lesson emails. We will then proceed with recording Gurudeva’s other books as well.
One challenge we face is the pronunciation of Sanskrit words mixed with English. The English-language AI model we use does not recognize diacritical marks or long and short vowels to pronounce Sanskrit words accurately. Fortunately, ElevenLabs allows us to upload a lexicon text file with word replacements. When the AI is reading a text block, it checks this file for any word spelling changes and then pronounces the corrected version. This is a slow process as we experiment with different spellings until the AI gets it right. We are currently building the lexicon.
We are dedicated to using this digital voice solely for reproducing Gurudeva’s original writings or speeches. Crossing this line would not only be misleading but also unfair to all parties involved, including Gurudeva himself.
Enough talk. Below is a preview of Gurudeva’s digital voice reading from Merging with Siva, lesson one. As the AI model improves, we anticipate even higher quality recordings in the future.