File size: 2,847 Bytes
aa9be1e |
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 |
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
Process songs.jsonl to generate corresponding lrc files and jsonl files.
"""
import json
import os
import re
from pathlib import Path
from typing import List
INPUT_JSONL = Path("xxx/diffrhythm2/example/final_zh_test.jsonl")
OUTPUT_SONG_DIR = Path("xxx/diffrhythm2/example/zh_songs")
OUTPUT_LRC_DIR = Path("xxx/diffrhythm2/example/zh_lrc")
TIMESTAMP_PATTERN = re.compile(r"\[\d{2}:\d{2}(?:\.\d+)?\]")
STRUCTURE_PATTERN = re.compile(r"^\[[^\]]+\]$")
def normalize_structure(tag: str) -> str:
"""Convert structure tag to target format."""
tag_lower = tag.lower()
if tag_lower.startswith("verse"):
return "[verse]"
if "chorus" in tag_lower:
return "[chorus]"
if "bridge" in tag_lower:
return "[bridge]"
return f"[{tag_lower}]"
def transform_lyrics(raw_lyrics: str) -> List[str]:
"""Convert lyrics to LRC line list according to requirements."""
lines = ["[start]", "[intro]"]
for raw_line in raw_lyrics.splitlines():
line = raw_line.strip()
if not line:
continue
# Process structure tags separately
if STRUCTURE_PATTERN.match(line) and not TIMESTAMP_PATTERN.match(line):
tag_content = line[1:-1].strip()
lines.append(normalize_structure(tag_content))
continue
# Remove timestamps
text = TIMESTAMP_PATTERN.sub("", line).strip()
if not text:
continue
lines.append(text)
lines.append("[end]")
return lines
def ensure_dirs() -> None:
OUTPUT_SONG_DIR.mkdir(parents=True, exist_ok=True)
OUTPUT_LRC_DIR.mkdir(parents=True, exist_ok=True)
def process_songs() -> None:
ensure_dirs()
with INPUT_JSONL.open("r", encoding="utf-8") as infile:
for idx, line in enumerate(infile, start=1):
line = line.strip()
if not line:
continue
data = json.loads(line)
description = data.get("description", "")
lyrics_raw = data.get("lyrics", "")
lrc_lines = transform_lyrics(lyrics_raw)
lrc_filename = f"song_{idx}.lrc"
lrc_path = OUTPUT_LRC_DIR / lrc_filename
lrc_path.write_text("\n".join(lrc_lines), encoding="utf-8")
song_base = f"song_{idx}"
song_filename = f"{song_base}.jsonl"
song_json_path = OUTPUT_SONG_DIR / song_filename
song_entry = {
"song_name": song_base,
"style_prompt": description,
"lyrics": f"example/zh_lrc/{lrc_filename}",
}
song_json_path.write_text(json.dumps(song_entry, ensure_ascii=False) + "\n", encoding="utf-8")
print(f"Processed song {idx}: {song_filename}")
if __name__ == "__main__":
process_songs()
|