Synthetic Voice Dataset¶

Emir discusses the work done in synthetic singing voice datasets by Matthew Awlett at Cereproc

• Speech Synthesis, as seen on https://www.youtube.com/watch?v=G-zyTlZQYpE
• Using the NUS sung and spoken labels corpus
• Using 'copy-synthesis' and parametric systems
• Question of what should be included in vocal singing datasets

• Whisper, spoken word, voice-like instruments, breath, elongated decay of words, reverberated tails