import numpy as np import streamlit as st import streamlit_pianoroll from fortepyan import MidiPiece st.set_page_config( page_title="PianoRoll Demo", page_icon=":musical_keyboard:", ) def main(): piano_music_demo() def piano_music_demo(): piece = MidiPiece.from_file("haydn.mid") # TODO Improve fortepyan to make this cleaner piece.time_shift(-piece.df.start.min()) st.write("## Display a PianoRoll player") st.write(""" The core functionality of pianorolls includes music playback and visualization. If you have a MIDI file with piano music, see here for instructions on interacting with it using Streamlit components. """) code = """ import streamlit_pianoroll from fortepyan import MidiPiece piece = MidiPiece.from_file("haydn.mid") streamlit_pianoroll.from_fortepyan(piece) """ st.code(code, language="python") streamlit_pianoroll.from_fortepyan(piece) st.info( body="This component is dedicated to piano music, there's no way to interract with multiple instruments.", icon="🎹", ) st.write("## Conditional coloring") st.write(""" To create a pianoroll with different notes in separate colors, create two `MidiPiece` objects, each containing the notes for one color. """) st.write("#### Absolute pitch value condition") st.write(""" Here's how to highlight notes with pitch above or below a certain threshold. Value of pitch 60 corresponds to the middle C (C4) on a piano keyboard ([refrence table](https://inspiredacoustics.com/en/MIDI_note_numbers_and_center_frequencies)). """) code = """ df = piece.df ids = df.pitch > pitch_threshold part_a = df[ids] part_b = df[~ids] piece_a = MidiPiece(df=part_a) piece_b = MidiPiece(df=part_b) streamlit_pianoroll.from_fortepyan( piece=piece_a, secondary_piece=piece_b, ) """ st.code(code, language="python") df = piece.df.copy() pitch_threshold = st.number_input( label="pitch_threshold", min_value=df.pitch.min(), max_value=df.pitch.max(), value=60, ) ids = df.pitch > pitch_threshold part_a = df[ids].copy() part_b = df[~ids].copy() piece_a = MidiPiece(df=part_a) piece_b = MidiPiece(df=part_b) streamlit_pianoroll.from_fortepyan( piece=piece_a, secondary_piece=piece_b, ) st.write("#### Note duration condition") st.write(""" Here's how to highlight notes based on their absolute duration, which can differentiate between fast and slow notes. """) code = """ df = piece.df ids = df.duration > duration_threshold part_a = df[ids] part_b = df[~ids] piece_a = MidiPiece(df=part_a) piece_b = MidiPiece(df=part_b) streamlit_pianoroll.from_fortepyan( piece=piece_a, secondary_piece=piece_b, ) """ st.code(code, language="python") df = piece.df.copy() duration_threshold = st.number_input( label="duration threshold [s]", min_value=0.05, max_value=5., value=0.25, ) ids = df.duration > duration_threshold part_a = df[ids].copy() part_b = df[~ids].copy() piece_a = MidiPiece(df=part_a) piece_b = MidiPiece(df=part_b) streamlit_pianoroll.from_fortepyan( piece=piece_a, secondary_piece=piece_b, ) st.write("#### Music generation prompting") st.write(""" Here's how to highlight show results of a generative algorithm that is supposed to work musically with an input from. We can take the notes with pitch below 72 (C5) as a prompt, and display it together with the generation. This sample performs random shuffle of the note pitch values, so the results are not muscially great (your algorithms should be better). """) df = piece.df.copy() ids = df.pitch < 72 part_a = df[ids].copy() piece_a = MidiPiece(df=part_a) # Fake random algorithm part_b = df[~ids].copy() part_b.pitch = np.random.permutation(part_b.pitch) part_b.velocity = np.random.permutation(part_b.velocity) piece_b = MidiPiece(df=part_b) streamlit_pianoroll.from_fortepyan( piece=piece_a, secondary_piece=piece_b, ) if __name__ == "__main__": main()