An Automated End-to-End Open-Source Software for High-Quality Text-to-Speech Dataset Generation

arXiv:2402.16380v1 Announce Type: cross
Abstract: Data availability is crucial for advancing artificial intelligence applications, including voice-based technologies. As content creation, particularly in social media, experiences increasing demand, translation and text-to-speech (TTS) technologies have become essential tools. Notably, the performance of these TTS technologies is highly dependent on the quality of the training data, emphasizing the mutual dependence of data availability and technological progress. This paper introduces an end-to-end tool to generate high-quality datasets for text-to-speech (TTS) models to address this …

abstract applications artificial artificial intelligence arxiv automated availability become cs.ai cs.cl cs.lg data dataset dataset generation demand eess.as intelligence media performance quality social social media software speech technologies text text-to-speech tools translation tts type voice