A list of demo websites for automatic music generation research
###interactive multi-track music composition
- Jamming with Yating (RNN; hsiao19ismir-lbd): https://www.youtube.com/watch?v=9ZIJrr6lmHg
###interactive piano composition
- Piano Genie (RNN; donahue18nips-creativity): https://piano-genie.glitch.me/
- AI duet (RNN; roberts16nips-demo): https://experiments.withgoogle.com/ai/ai-duet/view/
###compose melody
- SSMGAN (VAE+LSTM+GAN; jhamtani19ml4md): https://drive.google.com/drive/folders/1TlOrbYAm7vGUvRrxa-uiH17bP-4N4e9z
- StructureNet (LSTM; medeot18ismir) https://www.dropbox.com/sh/yxkxlnzi913ba50/AAA_mDbhdmaGJC9qj0zSlqCea?dl=0
- MusicVAE (LSTM+VAE; roberts18icml): https://magenta.tensorflow.org/music-vae
- MidiNet (CNN+GAN; yang17ismir): https://richardyang40148.github.io/TheBlog/midinet_arxiv_demo.html
- C-RNN-GAN (LSTM+GAN; mogren16cml): http://mogren.one/publications/2016/c-rnn-gan/
- folkRNN (LSTM): https://folkrnn.org/
###compose single-track polyphonic music
- PopRNN (RNN; yeh19ismir-lbd): https://soundcloud.com/yating_ai/sets/ismir-2019-submission/
- VGMIDI (LSTM; ferreira19ismir): N/A
- Modularized VAE (GRU+VAE; wang19icassp): https://github.com/MiuLab/MVAE_Music
- BachProp (GRU; colombo18arxiv): https://sites.google.com/view/bachprop
###compose multitrack music
- measure-by-measure (RNN): https://sites.google.com/view/pjgbjzom
- JazzRNN (RNN; yeh19ismir-lbd): https://soundcloud.com/yating_ai/sets/ismir-2019-submission/
- MIDI-Sandwich2 (RNN+VAE; liang19arxiv): https://github.com/LiangHsia/MIDI-S2
- LakhNES (Transformer; donahue19ismir): https://chrisdonahue.com/LakhNES/
- MuseNet (Transformer): https://openai.com/blog/musenet/
- MIDI-VAE (GRU+VAE; brunner18ismir): https://www.youtube.com/channel/UCCkFzSvCae8ySmKCCWM5Mpg
- Multitrack MusicVAE (LSTM+VAE; simon18ismir): https://magenta.tensorflow.org/multitrack
- MuseGAN (CNN+GAN; dong18aaai): https://salu133445.github.io/musegan/
###given chord, compose melody
- JazzGAN (GAN; trieu18mume): https://www.cs.hmc.edu/~keller/jazz/improvisor/
- XiaoIce Band (GRU; zhu18kdd): http://tv.cctv.com/2017/11/24/VIDEo7JWp0u0oWRmPbM4uCBt171124.shtml
###given lyrics, compose melody
- Conditional LSTM-GAN (LSTM+GAN; yu19arxiv): https://github.com/yy1lab/Lyrics-Conditioned-Neural-Melody-Generation
- iComposer (LSTM; lee19acl): https://www.youtube.com/watch?v=Gstzqls2f4A
- SongWriter (GRU; bao18arxiv): N/A
###compose drum loops
- Wei's model (VAE+GAN; wei19ismir): https://github.com/Sma1033/drum_generation_with_ssm
- DrumNet (GAE; lattner19waspaa): https://sites.google.com/view/drum-generation
- Generating structured drum pattern using variational autoencoder and self-similarity matrix (wei19ismir): N/A
- DrumVAE (GRU+VAE; thio19milc): http://vibertthio.com/drum-vae-client
###compose melody+chords (two tracks)
- LeadsheetGAN (CRNN+GAN; liu18icmla): https://liuhaumin.github.io/LeadsheetArrangement/results
- LeadsheetVAE (RNN+VAE; liu18ismir-lbd): https://liuhaumin.github.io/LeadsheetArrangement/results
###given melody, compose arrangement
- LeadsheetGAN: see above
- LeadsheetVAE: see above
- XiaoIce Band (the "multi-instrument co-arrangement model"): N/A
###given prime melody, compose melody+chords
- local_conv_music_generation (CNN; ouyang18arxiv): https://somedaywilldo.github.io/local_conv_music_generation/
###given prime melody, compose melody+chords+bass
- BandNet (RNN; zhou18arxiv): https://soundcloud.com/yichao-zhou-555747812/sets/bandnet-sound-samples-1
###given piano score, compose an orchestration
- LOP (RBM; crestel17smc): https://qsdfo.github.io/LOP/results.html
###symbolic-domain genre style transfer
- Pop2Jazz (RNN; yeh19ismir-lbd): https://soundcloud.com/yating_ai/sets/ismir-2019-submission/
- supervised (RNN; cífka19ismir): https://github.com/cifkao/ismir2019-music-style-translation
- CycleGAN2 (CNN+GAN; brunner19mml): https://drive.google.com/drive/folders/1Jr_p6pnKvhA2YW9sp-ABChiFgV3gY1aT
- CycleGAN (CNN+GAN; brunner18ictai): https://github.com/sumuzhao/CycleGAN-Music-Style-Transfer
- FusionGAN (GAN; chen17icdm): http://people.cs.vt.edu/czq/publication/fusiongan/
###symbolic-domain arrangement style transfer
- UnetED (CNN+Unet; hung19ijcai): https://biboamy.github.io/disentangle_demo/result/index.html
###symbolic-domain rhythm style transfer
###given MIDI, generate human-like MIDI: Piano only
- GGNN (graph NN + hierarchical attention RNN; jeong19icml)
- VirtuosoNet (LSTM+hierarchical attention network; jeong18nipsw): https://www.youtube.com/playlist?list=PLkIVXCxCZ08rD1PXbrb0KNOSYVh5Pvg-c
- Wave2Midi2Wave (transformer+wavenet; hawthorne19iclr): https://magenta.tensorflow.org/maestro-wave2midi2wave
- PerformanceRNN (RNN): https://magenta.tensorflow.org/performance-rnn
###given MIDI, generate human-like MIDI: Drum only
- GrooVAE (seq2seq+VAE; gillick19icml): https://magenta.tensorflow.org/groovae
###given score, generate musical audio (performance): Not limited to Piano
- PerformanceNet (CNN+GAN; wang19aaai): https://github.com/bwang514/PerformanceNet
- Conditioned Wavenet (Wavenet; manzelli18ismir): http://people.bu.edu/bkulis/projects/music/index.html
###audio synthesis
- DDSP (; lamtharn20iclr): https://storage.googleapis.com/ddsp/index.html
- MelNet (auto-regressive; vasquez19arxiv): https://audio-samples.github.io/
- AdVoc (; neekhara19arxiv): http://chrisdonahue.com/advoc_examples/
- GANSynth (CNN+GAN; engel19iclr): https://magenta.tensorflow.org/gansynth
- SynthNet (schimbinschi19ijcai): https://www.dropbox.com/sh/hkp3o5xjyexp2x0/AADvrfXTbHBXs9W7GN6Yeorua?dl=0
- TiFGAN (CNN+GAN; marafioti19arxiv): https://tifgan.github.io/
- SING (defossez18nips): https://research.fb.com/wp-content/themes/fb-research/research/sing-paper/
- WaveGAN (CNN+GAN; donahue19iclr): https://github.com/chrisdonahue/wavegan
- NSynth (WaveNet; engel17arxiv): https://magenta.tensorflow.org/nsynth
###audio-domain music generation
- dadabots (sampleRNN; carr18mume): http://dadabots.com/music.php
###audio-domain singing synthesis
- mellotron: https://nv-adlr.github.io/Mellotron
- lee's model (lee19arxiv): http://ksinging.mystrikingly.com/
- http://home.ustc.edu.cn/~yiyh/interspeech2019/
###audio-domain singing style transfer
- SINGAN (GAN; sisman19apsipa): N/A
- [MSVC-GAN] (GAN): https://hujinsen.github.io/
- https://mtg.github.io/singing-synthesis-demos/voice-cloning/
- https://enk100.github.io/Unsupervised_Singing_Voice_Conversion/
- Yong&Nam (DSP; yong18icassp): https://seyong92.github.io/singing-expression-transfer/
- cybegan (CNN+GAN; wu18faim): http://mirlab.org/users/haley.wu/cybegan/
###audio-domain singing correction
- deep-autotuner (CGRU; wagner19icassp): http://homes.sice.indiana.edu/scwager/deepautotuner.html
###audio-domain style transfer (general)
- MelGAN-VC (GAN; pasini19arxiv): https://www.youtube.com/watch?v=3BN577LK62Y&feature=youtu.be
- RaGAN (GAN; lu19aaai): https://github.com/ChienYuLu/Play-As-You-Like-Timbre-Enhanced-Multi-modal-Music-Style-Transfer
- TimbreTron (GAN; huang19iclr): https://www.cs.toronto.edu/~huang/TimbreTron/samples_page.html
- string2woodwind (DSP; wagner17icassp): http://homes.sice.indiana.edu/scwager/css.html