Nowadays, voice cloning technology has advanced to its fullest potential and is able to generate highly realistic synthetic human voices. There are several alternatives that people can do who want to get with voice cloning without paying anything.
To begin with, voice cloning exists in the open-sourced tools that Mozilla and others have released. Mozilla must be trained with some data, and their publicly available TTS requires the user to train a model from scratch using this dataset. This training is priced on the basis of how big & good dataset you use. The cost of computing resources, for smaller projects: $100 to $500 per month or a similar period.
Secondly, there are various platforms like DUPDUB that offer instant voice cloning services and some of them are even free. Thanks to the free tier you will be given a couple minutes of voice data which is great for anyone curious about getting into this. The free version of the service usually only supports simple features — and getting access to advanced settings normally requires paying a fee.
Finally, educational purposes and companies and academic institutions who research voice cloning free access to this repository. Google Tacotron 2 Google recently released the state of art neural network for text-to-speech synthesis named as “tacotron”, these kinds of tool enable to explore voice cloning. The Google models, from results available via academic partnerships, through to further understanding without the expense of actual research.
In addition, deep learning enabled music and voice generation tools (for example OpenAI's Jukebox) have been used in research context as well. OpenAI has made its models available via research collaborations and partnerships, with the price of access varying based on use case and projects scale.
With the increase in technology advancements, there are a growing number of free tools to change your voice giving users endless combinations for all types of voices.