XinleiRen / sing-segmentation-evaluation

F0-estimate and onset detect for singing

• the project aims to solve F0 estimate and onset detect using mfshs algorithm and CNN for singing.
• the Application scenario include solfege、 Karaoke and various Monophonic scenes

include algorithm

1. F0 estimate (similar to yin)
2. sw alignment
3. CNN for onset detect
4. post process by combining the above algorithm

requirement

• tensorflow
• numpy
• librosa

run

python main.py

onset detection method cite this paper

@conference {rgong2018singing,
	title = {Singing voice phoneme segmentation by hierarchically inferring syllable and phoneme onset positions},
	booktitle = {Interspeech 2018},
	year = {2018},
	month = {02/09/2018},
	address = {Hyderabad, India},
	url = {https://arxiv.org/pdf/1806.01665.pdf},
	author = {Rong Gong and Xavier Serra}
}

result