Deep Learning for Jazz Walking Bass Transcription

Abeßer J, Balke S, Frieler K, Pfleiderer M, Müller M. Deep Learning for Jazz Walking Bass Transcription. In: AES Conference: 2017 AES International Conference on Semantic Audio. Audio Engineering Society; 2017. Paper 5-2. Available from: https://aes.org/publications/elibrary-page/?id=18762

@inproceedings{Abeser2017_18762,
  author = {Abeßer, Jakob and Balke, Stefan and Frieler, Klaus and Pfleiderer, Martin and Müller, Meinard},
  title = {{Deep Learning for Jazz Walking Bass Transcription}},
  booktitle = {AES Conference: 2017 AES International Conference on Semantic Audio},
  note = {Paper 5-2},
  year = {2017},
  month = jun,
  publisher = {Audio Engineering Society},
  url = {https://aes.org/publications/elibrary-page/?id=18762}
}

TY  - CPAPER
TI  - Deep Learning for Jazz Walking Bass Transcription
AU  - Abeßer, Jakob
AU  - Balke, Stefan
AU  - Frieler, Klaus
AU  - Pfleiderer, Martin
AU  - Müller, Meinard
T2  - AES Conference: 2017 AES International Conference on Semantic Audio
M1  - Paper 5-2
PY  - 2017
DA  - 2017/06/06
UR  - https://aes.org/publications/elibrary-page/?id=18762
PB  - Audio Engineering Society
LA  - en
AB  - In this paper, we focus on transcribing walking bass lines, which provide clues for revealing the actual played chords in jazz recordings. Our transcription method is based on a deep neural network (DNN) that learns a mapping from a mixture spectrogram to a salience representation that emphasizes the bass line. Furthermore, using beat positions, we apply a late-fusion approach to obtain beat-wise pitch estimates of the bass line. First, our results show that this DNN-based transcription approach outperforms state-of-the-art transcription methods for the given task. Second, we found that an augmentation of the training set using pitch shifting improves the model performance. Finally, we present a semi-supervised learning approach where additional training data is generated from predictions on unlabeled datasets.
ER  -