Mixing Synthetic and Recorded Signals for Audio-Book Generation

• Published in: Speech and Computer Vol. 12335; pp. 479 - 489
• Main Authors: Shamsi, Meysam, Barbot, Nelly, Lolive, Damien, Chevelu, Jonathan
• Format: Book Chapter
• Language: English
• Published: Switzerland Springer International Publishing AG 2020; Springer International Publishing
• Series: Lecture Notes in Computer Science
• Subjects: Audio-book generation; Quality evaluation; Text-to-Speech
• ISBN: 3030602753; 9783030602758
• ISSN: 0302-9743; 1611-3349
• DOI: 10.1007/978-3-030-60276-5_46

Using TTS systems helps to reduce the cost of audio-book generation. This paper investigates the idea of mixing synthetic and recorded natural speech signals to control the trade-off between the overall quality of audio book and its production cost. Firstly, fully synthetic signals and mixed synthetic and natural signals are compared perceptually using different levels of synthetic quality. The listeners’ perception shows that mixed signals are preferred. Next, the order and configuration of mixed signals are studied. The perceptual test does not show any significant difference between the different configurations. Finally, the synthetic quality and the bias of a starting and ending part of mixed signals in perceptual test are investigated.