Spaces:
Sleeping
Sleeping
File size: 4,299 Bytes
91394e0 bbfbc08 91394e0 7a23964 91394e0 7a23964 91394e0 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 |
import random
from typing import Iterator
from data_handlers import get_melody_handler
from .utils.g2p import preprocess_text
class MelodyController:
def __init__(self, melody_source_id: str, cache_dir: str):
self.melody_source_id = melody_source_id
self.song_id = None
# load song database if needed
parts = self.melody_source_id.split("-")
self.mode = parts[0]
self.align_type = parts[1]
dataset_name = parts[-1]
if dataset_name == "none":
self.database = None
else:
handler_cls = get_melody_handler(dataset_name)
self.database = handler_cls(self.align_type, cache_dir)
def get_melody_constraints(self, max_num_phrases: int = 5) -> str:
"""Return a lyric-format prompt based on melody structure."""
if self.mode == "gen":
return ""
elif self.mode == "sample":
assert self.database is not None, "Song database is not loaded."
self.song_id = random.choice(self.database.get_song_ids())
self.reference_song = self.database.iter_song_phrases(self.song_id)
phrase_length = self.database.get_phrase_length(self.song_id)
if not phrase_length:
return ""
prompt = (
"\n请按照歌词格式回复，每句需遵循以下字数规则："
+ "".join(
[
f"\n第{i}句：{c}个字"
for i, c in enumerate(phrase_length[:max_num_phrases], 1)
]
)
+ "\n如果没有足够的信息回答，请使用最少的句子，不要重复、不要扩展、不要加入无关内容。\n"
)
return prompt
else:
raise ValueError(f"Unsupported melody mode: {self.mode}")
def generate_score(
self, lyrics: str, language: str
) -> list[tuple[float, float, str, int]]:
"""
lyrics: [lyric, ...]
returns: [(start, end, lyric, pitch), ...]
"""
text_list = preprocess_text(lyrics, language)
if self.mode == "gen" and self.align_type == "random":
return self._generate_random_score(text_list)
elif self.mode == "sample":
if not self.reference_song:
raise RuntimeError(
"Must call get_melody_constraints() before generate_score() in sample mode."
)
return self._align_text_to_score(
text_list, self.reference_song, self.align_type
)
else:
raise ValueError(f"Unsupported melody_source_id: {self.melody_source_id}")
def _generate_random_score(self, text_list: list[str]):
st = 0
score = []
for lyric in text_list:
pitch = random.randint(57, 69)
duration = round(random.uniform(0.1, 0.5), 4)
ed = st + duration
score.append((st, ed, lyric, pitch))
st = ed
return score
def _align_text_to_score(
self,
text_list: list[str],
song_phrase_iterator: Iterator[dict],
align_type: str,
):
score = []
text_idx = 0
while text_idx < len(text_list):
reference = next(song_phrase_iterator)
for st, ed, ref_lyric, pitch in zip(
reference["note_start_times"],
reference["note_end_times"],
reference["note_lyrics"],
reference["note_midi"],
):
assert ref_lyric not in [
"<AP>",
"<SP>",
], f"Proccessed {self.melody_source_id} score segments should not contain <AP> or <SP>." # TODO: remove in PR, only for debug
if pitch == 0:
score.append((st, ed, ref_lyric, pitch))
elif ref_lyric in ["-", "——"] and align_type == "lyric":
score.append((st, ed, "-", pitch))
else:
score.append((st, ed, text_list[text_idx], pitch))
text_idx += 1
if text_idx >= len(text_list):
break
return score
|