becaliang/Music_Generation_Project

1from music21 import converter, instrument, key, note, stream, tempo, meter2from collections import OrderedDict3from itertools import groupby4from grammar import parse_melody5import copy6 7# ---------------------------- INTERNAL HELPERS ---------------------------- #8 9def _load_and_extract(data_path):10    midi_data = converter.parse(data_path)11    parts = midi_data.parts.stream()12    solo_parts = []13 14    for p in parts:15        voice = stream.Voice()16        for el in p.recurse():17            if isinstance(el, (note.Note, note.Rest)):18                if el.quarterLength == 0.0:19                    el.quarterLength = 0.2520                voice.insert(el.offset, el)21            elif isinstance(el, (instrument.Instrument, tempo.MetronomeMark, key.KeySignature, meter.TimeSignature)):22                voice.insert(0, el)23        solo_parts.append(voice)24 25    return solo_parts26 27def _split_into_measures(tracks):28    melody_by_track = {}29    chords_by_track = {}30 31    for i, voice_stream in enumerate(tracks):32        track_name = f"Track_{i + 1}"33        melody_measures = OrderedDict()34        melody_offsets = [(int(n.offset / 4), n) for n in voice_stream if isinstance(n, (note.Note, note.Rest))]35 36        for idx, group in groupby(melody_offsets, lambda x: x[0]):37            melody_measures[idx] = [n[1] for n in group]38 39        chord_stream = copy.deepcopy(voice_stream)40        chord_stream.removeByClass(note.Note)41        chord_stream.removeByClass(note.Rest)42        chord_offsets = [(int(n.offset / 4), n) for n in chord_stream]43 44        chords = OrderedDict()45        for idx, group in groupby(chord_offsets, lambda x: x[0]):46            chords[idx] = [n[1] for n in group]47 48        min_len = min(len(chords), len(melody_measures))49        chords = OrderedDict(list(chords.items())[:min_len])50        melody_measures = OrderedDict(list(melody_measures.items())[:min_len])51 52        melody_by_track[track_name] = melody_measures53        chords_by_track[track_name] = chords54 55    return melody_by_track, chords_by_track56 57def _encode_grammars(melody_dict, chord_dict):58    grammars_by_track = {}59 60    for track_name in melody_dict:61        melody_measures = melody_dict.get(track_name)62        chord_measures = chord_dict.get(track_name)63 64        if not isinstance(melody_measures, dict) or not isinstance(chord_measures, dict):65            print(f"⚠️ Skipping {track_name}: Invalid measure format.")66            continue67 68        grammar_list = []69        for idx in sorted(melody_measures.keys()):70            if idx not in chord_measures:71                continue72 73            m_stream = stream.Voice()74            c_stream = stream.Voice()75 76            for n in melody_measures[idx]:77                m_stream.insert(n.offset, n)78            for c in chord_measures[idx]:79                c_stream.insert(c.offset, c)80 81            grammar = parse_melody(m_stream, c_stream)82            grammar_list.append(grammar)83 84        grammars_by_track[track_name] = grammar_list85 86    return grammars_by_track87 88# ---------------------------- PUBLIC INTERFACE ---------------------------- #89 90def get_multi_track_musical_data(data_path):91    solo_parts = _load_and_extract(data_path)92    melody_dict, chord_dict = _split_into_measures(solo_parts)93    grammar_dict = _encode_grammars(melody_dict, chord_dict)94 95    merged_chords = {}96    for chords in chord_dict.values():97        for k, v in chords.items():98            if k not in merged_chords:99                merged_chords[k] = []100            merged_chords[k].extend(v)101 102    return merged_chords, grammar_dict103 104def get_multi_track_corpus_data(grammar_dict):105    corpus_dict = {}106    all_tokens = []107 108    for track, grammar_list in grammar_dict.items():109        tokens = [token for phrase in grammar_list for token in phrase.split(" ")]110        corpus_dict[track] = tokens111        all_tokens.extend(tokens)112 113    vocab = sorted(set(all_tokens))114    token_to_index = {t: i for i, t in enumerate(vocab)}115    index_to_token = {i: t for t, i in token_to_index.items()}116 117    return corpus_dict, vocab, token_to_index, index_to_token118 119def get_musical_data(data_path):120    solo_parts = _load_and_extract(data_path)121    melody_dict, chords_dict = _split_into_measures(solo_parts)122    grammar_dict = _encode_grammars(melody_dict, chords_dict)123 124    merged_chords = {}125    for chords in chords_dict.values():126        for k, v in chords.items():127            if k not in merged_chords:128                merged_chords[k] = []129            merged_chords[k].extend(v)130 131    all_grammars = [g for track_grammars in grammar_dict.values() for g in track_grammars]132    return merged_chords, all_grammars133 134def get_corpus_data(grammar_list):135    corpus = [token for phrase in grammar_list for token in phrase.split()]136    vocab = sorted(set(corpus))137    token_to_index = {t: i for i, t in enumerate(vocab)}138    index_to_token = {i: t for i, t in enumerate(vocab)}139    return corpus, vocab, token_to_index, index_to_token140