Yes, it is possible to program audio to synthesize a human-like voice. This can be done through the use of text-to-speech AI models, which use machine learning algorithms to convert text input into spoken words. One such example is the new AI model from Microsoft, called VALLE. This model is able to closely simulate a person's voice when given a three-second audio sample. The system works by learning the discrete codes derived from an off-the-shelf neural audio codec model, as well as 60,000 hours of speech, which is 100 times more than existing text-to-speech systems.
Another example is the Canadian AI startup named Lyrebird, which has developed algorithms that can clone anyone's voice by listening to just a single minute of sample audio. The company's algorithms can then generate synthetic speech that closely mimics the person's natural voice. Additionally, there are other AI-powered voice generators such as PlayHT, ElevenLabs, and MurfAI that can create human-like, realistic audio from the text you input using various AI-powered voice models. These advancements in AI technology have made it possible to generate synthetic speech that closely mimics natural human speech, making it difficult for the human ear to distinguish between the two. However, it is important to note that there are also researchers working on ways to detect deepfake audio, as it can be used for malicious purposes.