Pop2Piano: A Transformer Network for Generating Piano covers from waveforms of pop music

Home

Discover

Bookmarks

Keep Up With Latest Trending Papers. Computer Science, AI and Machine Learning and more.Subscribe

Audio and Speech Processing

Pop2Piano: A Transformer Network for Generating Piano covers from waveforms of pop music

Pop2Piano : Pop Audio-based Piano Cover Generation

A preprocessing pipeline for synchronizing and filtering paired

We present a model that generates a piano cover from given waveforms of pop music without melodyand chord extraction modules.

We make a large amount of paired and synchronized {pop, piano cover} data using an automated pipeline and show that our model can generate plausible piano covers.

This is the first model to directly generate a piano cover from pop audio without melodyand chord extraction modules.

Authors

Jongho Choi, Kyogu Lee

Related Topics

Deep learning

Preprocessing pipeline

Intensive deep learning-based methods

Pop2piano

Melody and chord extraction modules

Transformer-based piano cover generation model pop2piano

Preprocessing system

Melody or chord extractor

Melody or chord extraction modules

Introduction

In this study, we introduce a study on the transformer-based piano cover generation model pop2piano, which generates a piano performance (midi) from the waveform of pop music directly.

We build 300-hour of synchronized dataset, called iano cover ynchronized to op audio (psp), and introduce the preprocessing system used to make this dataset.

A list of data and preprocessing codes to reproduce the psp dataset, and release an executable pop2piano demo on colab.

Result

We present a novel study on generating pop piano covers directly from audio without using melody or chord extraction modules based on a transformer network.

We collect 300 hours of paired datasets to train the model and we design a pipeline that synchronizes them in a form suitable for training neural networks.

Then we show and evaluate that the model can generate plausible pop piano covers and also can mimic the style of a specific arranger.

Read the Paper

Read Paper

Bookmarks

Bookmarks

Keep Up With Latest Trending Papers. Computer Science, AI and Machine Learning and more.Subscribe

Audio and Speech Processing

Pop2Piano: A Transformer Network for Generating Piano covers from waveforms of pop music

Pop2Piano : Pop Audio-based Piano Cover Generation

A preprocessing pipeline for synchronizing and filtering paired

We present a model that generates a piano cover from given waveforms of pop music without melodyand chord extraction modules.

We make a large amount of paired and synchronized {pop, piano cover} data using an automated pipeline and show that our model can generate plausible piano covers.

This is the first model to directly generate a piano cover from pop audio without melodyand chord extraction modules.

Authors

Jongho Choi, Kyogu Lee

Related Topics

Deep learning

Preprocessing pipeline

Intensive deep learning-based methods

Pop2piano

Melody and chord extraction modules

Transformer-based piano cover generation model pop2piano

Preprocessing system

Melody or chord extractor

Melody or chord extraction modules

Introduction

In this study, we introduce a study on the transformer-based piano cover generation model pop2piano, which generates a piano performance (midi) from the waveform of pop music directly.

We build 300-hour of synchronized dataset, called iano cover ynchronized to op audio (psp), and introduce the preprocessing system used to make this dataset.

A list of data and preprocessing codes to reproduce the psp dataset, and release an executable pop2piano demo on colab.

Result

We present a novel study on generating pop piano covers directly from audio without using melody or chord extraction modules based on a transformer network.

We collect 300 hours of paired datasets to train the model and we design a pipeline that synchronizes them in a form suitable for training neural networks.

Then we show and evaluate that the model can generate plausible pop piano covers and also can mimic the style of a specific arranger.

Read the Paper

Read Paper

Home

Discover

Bookmarks

◐ Recommended Follows

◐ Latest Activity