pop2piano

Pop2Piano is an encoder-decoder Transformer model based on T5. The input audio is transformed to its waveform and passed to the encoder, which transforms it to a latent representation. The decoder uses these latent representations to generate token ids in an autoregressive way. Each token id corresponds to one of four different token types: time, velocity, note and 'special'. The token ids are then decoded to their equivalent MIDI file.

Model Sources

• Paper
• Original Repository
• HuggingFace Space Demo

Usage

To use Pop2Piano, you will need to install the 🤗 Transformers library, as well as the following third party modules:

pip install git+https://github.com/huggingface/transformers.git
pip install pretty-midi==0.2.9 essentia==2.1b6.dev1034 librosa scipy

Please note that you may need to restart your runtime after installation.

Pop music to Piano

Code Example

• Using your own Audio

>>> import librosa
>>> from transformers import Pop2PianoForConditionalGeneration, Pop2PianoProcessor

>>> audio, sr = librosa.load("", sr=44100)  # feel free to change the sr to a suitable value.
>>> model = Pop2PianoForConditionalGeneration.from_pretrained("sweetcocoa/pop2piano")
>>> processor = Pop2PianoProcessor.from_pretrained("sweetcocoa/pop2piano")

>>> inputs = processor(audio=audio, sampling_rate=sr, return_tensors="pt")
>>> model_output = model.generate(input_features=inputs["input_features"], composer="composer1")
>>> tokenizer_output = processor.batch_decode(
...     token_ids=model_output, feature_extractor_output=inputs
... )["pretty_midi_objects"][0]
>>> tokenizer_output.write("./Outputs/midi_output.mid")

• Audio from Hugging Face Hub

>>> from datasets import load_dataset
>>> from transformers import Pop2PianoForConditionalGeneration, Pop2PianoProcessor

>>> model = Pop2PianoForConditionalGeneration.from_pretrained("sweetcocoa/pop2piano")
>>> processor = Pop2PianoProcessor.from_pretrained("sweetcocoa/pop2piano")
>>> ds = load_dataset("sweetcocoa/pop2piano_ci", split="test")

>>> inputs = processor(
...     audio=ds["audio"][0]["array"], sampling_rate=ds["audio"][0]["sampling_rate"], return_tensors="pt"
... )
>>> model_output = model.generate(input_features=inputs["input_features"], composer="composer1")
>>> tokenizer_output = processor.batch_decode(
...     token_ids=model_output, feature_extractor_output=inputs
... )["pretty_midi_objects"][0]
>>> tokenizer_output.write("./Outputs/midi_output.mid")

Example

Here we present an example of generated MIDI.

• Actual Pop Music

Your browser does not support the audio element.

• Generated MIDI

Your browser does not support the audio element.

Tips

1. Pop2Piano is an Encoder-Decoder based model like T5.
2. Pop2Piano can be used to generate midi-audio files for a given audio sequence.
3. Choosing different composers in Pop2PianoForConditionalGeneration.generate() can lead to variety of different results.
4. Setting the sampling rate to 44.1 kHz when loading the audio file can give good performance.
5. Though Pop2Piano was mainly trained on Korean Pop music, it also does pretty well on other Western Pop or Hip Hop songs.

Citation

BibTeX:

@misc{choi2023pop2piano,
      title={Pop2Piano : Pop Audio-based Piano Cover Generation},
      author={Jongho Choi and Kyogu Lee},
      year={2023},
      eprint={2211.00895},
      archivePrefix={arXiv},
      primaryClass={cs.SD}
}