jads/AudioTranscriberCombine
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

improve GUI layout and LRC stage controls

Browse Source 
- make the main page scrollable and keep the console reachable on small screens
- add fullscreen toggle (button + F11, Escape to exit)
- add enable switches for LRC generation/embedding and skip ready text.lrc
- sanitize multi-line ASR text before writing LRC timestamps
- emit LRC/embed progress once per file instead of twice
- increase progress-bar contrast
This commit is contained in:
jadssss
2026-10-27 22:29:13 +03:00
parent 15c0ba5d66
commit 999934f174
6 changed files with 135 additions and 24 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
.gitignore vendored
View File

@@ -14,3 +14,4 @@ output/
preset.json
*.log
combined_files.txt

3
core/config.py
View File

@@ -66,6 +66,9 @@ class LyricsConfig:
@dataclass
class LrcConfig:
enabled: bool = True # LRC generation stage
embed_enabled: bool = True # LRC embedding stage
skip_existing: bool = True # skip tracks with a ready text.lrc
language: Optional[str] = "ru"
align_model: Optional[str] = None
device: str = "auto"

4
core/lrc.py
View File

@@ -37,6 +37,10 @@ def segments_to_lrc(segments: list[dict]) -> str:
text = (seg.get("text") or "").strip()
if not text:
continue
# WhisperX can return multi-line text on long segments. An LRC line
# must be a single line: any newline would produce a broken line
# without a timestamp. Collapse all whitespace into single spaces.
text = re.sub(r"\s+", " ", text)
start = float(seg.get("start", 0.0))
lines.append(f"{format_timestamp(start)}{text}")
return "\n".join(lines) + "\n"

45
core/lrc_pipeline.py
View File

@@ -11,7 +11,9 @@ from .config import PipelineConfig
from .convert import convert_to_wav
from .lrc import embed_lrc, is_lrc, read_embedded_lyrics, segments_to_lrc
from .metadata import read_metadata
from .pipeline import FileResult, _save_unrecognized, collect_audio_files
from .pipeline import (
FileResult, _artifact_belongs_to, _save_unrecognized, collect_audio_files,
)
from .transcribe import transcribe_aligned
@@ -45,6 +47,20 @@ def _pick_audio_source(
return target
def _ready_lrc(track_out: Path, source: Path) -> Optional[str]:
"""Return existing valid LRC text for this exact source track, if any."""
if not _artifact_belongs_to(track_out, source):
return None
lrc_path = track_out / "text.lrc"
if not lrc_path.is_file():
return None
try:
content = lrc_path.read_text(encoding="utf-8")
except OSError:
return None
return content if content.strip() and is_lrc(content) else None
def run_lrc_pipeline(
config: PipelineConfig,
log: LogFn = print,
@@ -58,6 +74,10 @@ def run_lrc_pipeline(
output_dir = Path(config.output_dir)
output_dir.mkdir(parents=True, exist_ok=True)
if not config.lrc.enabled:
log("LRC: генерация отключена — пропуск этапа")
return []
files = collect_audio_files(input_dir, config.recursive)
if not files:
log("Аудиофайлы не найдены.")
@@ -69,6 +89,8 @@ def run_lrc_pipeline(
results: list[FileResult] = []
total = len(files)
if on_progress:
on_progress("lrc", 0, total)
if on_overall_progress:
on_overall_progress(0, total)
@@ -76,8 +98,6 @@ def run_lrc_pipeline(
if stop_event is not None and stop_event.is_set():
log("⏹ Отменено")
break
if on_progress:
on_progress("lrc", idx - 1, total)
log(f"\n▶ [{idx}/{total}] {f.name}")
track_out = output_dir / f.stem
@@ -85,6 +105,17 @@ def run_lrc_pipeline(
lrc_path = track_out / "text.lrc"
error: Optional[str] = None
if config.lrc.skip_existing:
ready = _ready_lrc(track_out, f)
if ready is not None:
log(" ⊘ text.lrc уже готов — пропуск")
results.append(FileResult(f, None, track_out, None, "existing+lrc"))
if on_progress:
on_progress("lrc", idx, total)
if on_overall_progress:
on_overall_progress(idx, total)
continue
tmp = Path(tempfile.mkdtemp(prefix=f"at_lrc_{f.stem}_"))
try:
embedded = read_embedded_lyrics(f) if config.lyrics.use_embedded_lyrics else None
@@ -151,6 +182,10 @@ def run_embed_pipeline(
output_dir = Path(config.output_dir)
output_dir.mkdir(parents=True, exist_ok=True)
if not config.lrc.embed_enabled:
log("LRC embed: встраивание отключено — пропуск этапа")
return []
files = collect_audio_files(input_dir, config.recursive)
if not files:
log("Аудиофайлы не найдены.")
@@ -161,6 +196,8 @@ def run_embed_pipeline(
results: list[FileResult] = []
total = len(files)
if on_progress:
on_progress("embed", 0, total)
if on_overall_progress:
on_overall_progress(0, total)
@@ -168,8 +205,6 @@ def run_embed_pipeline(
if stop_event is not None and stop_event.is_set():
log("⏹ Отменено")
break
if on_progress:
on_progress("embed", idx - 1, total)
lrc_path = output_dir / f.stem / "text.lrc"
error: Optional[str] = None

80
gui/main_window.py
View File

@@ -6,12 +6,12 @@ from pathlib import Path
from typing import Optional
from PySide6.QtCore import Qt, QUrl
from PySide6.QtGui import QColor, QDesktopServices
from PySide6.QtGui import QColor, QDesktopServices, QKeySequence, QShortcut
from PySide6.QtWidgets import (
QCheckBox, QComboBox, QDoubleSpinBox, QFileDialog, QFormLayout, QGroupBox,
QHBoxLayout, QLabel, QLineEdit, QMainWindow, QMessageBox,
QPlainTextEdit, QProgressBar, QPushButton, QSpinBox, QTabWidget,
QVBoxLayout, QWidget,
QCheckBox, QComboBox, QDoubleSpinBox, QFileDialog, QFormLayout, QFrame,
QGroupBox, QHBoxLayout, QLabel, QLineEdit, QMainWindow, QMessageBox,
QPlainTextEdit, QProgressBar, QPushButton, QScrollArea, QSpinBox,
QTabWidget, QVBoxLayout, QWidget,
)
from core.config import (
@@ -71,8 +71,17 @@ class MainWindow(QMainWindow):
# ---------------- UI ----------------
def _build_ui(self) -> None:
# The whole page lives in a scroll area: on small screens the user can
# scroll down to the log instead of the console being cut off by the
# bottom edge of the window.
scroll = QScrollArea()
scroll.setWidgetResizable(True)
scroll.setFrameShape(QFrame.Shape.NoFrame)
scroll.setHorizontalScrollBarPolicy(Qt.ScrollBarPolicy.ScrollBarAlwaysOff)
self.setCentralWidget(scroll)
central = QWidget()
self.setCentralWidget(central)
scroll.setWidget(central)
root = QVBoxLayout(central)
root.setContentsMargins(18, 16, 18, 16)
root.setSpacing(12)
@@ -82,6 +91,10 @@ class MainWindow(QMainWindow):
title.setObjectName("appTitle")
header.addWidget(title)
header.addStretch(1)
self.fullscreen_btn = QPushButton("⛶ На весь экран")
self.fullscreen_btn.setToolTip("Развернуть окно на весь экран (F11)")
self.fullscreen_btn.clicked.connect(self._toggle_fullscreen)
header.addWidget(self.fullscreen_btn)
self.device_label = QLabel("")
self.device_label.setObjectName("deviceInfo")
header.addWidget(self.device_label)
@@ -153,6 +166,7 @@ class MainWindow(QMainWindow):
self.convert_enabled_cb.toggled.connect(self._sync_tab_indicators)
self.separate_enabled_cb.toggled.connect(self._sync_tab_indicators)
self.transcribe_enabled_cb.toggled.connect(self._sync_tab_indicators)
self.lrc_enabled_cb.toggled.connect(self._sync_tab_indicators)
self.engine_cb.currentTextChanged.connect(self._sync_engine_fields)
row = QHBoxLayout()
@@ -191,9 +205,25 @@ class MainWindow(QMainWindow):
self.log_view = QPlainTextEdit()
self.log_view.setObjectName("log")
self.log_view.setReadOnly(True)
self.log_view.setMinimumHeight(240)
self.log_view.setPlaceholderText("Здесь появится лог выполнения…")
root.addWidget(self.log_view, 1)
self._fs_shortcut = QShortcut(QKeySequence("F11"), self)
self._fs_shortcut.activated.connect(self._toggle_fullscreen)
self._esc_shortcut = QShortcut(QKeySequence("Escape"), self)
self._esc_shortcut.activated.connect(self._leave_fullscreen)
def _toggle_fullscreen(self) -> None:
if self.isFullScreen():
self.showNormal()
else:
self.showFullScreen()
def _leave_fullscreen(self) -> None:
if self.isFullScreen():
self.showNormal()
def _with_browse(self, edit: QLineEdit, picker) -> QWidget:
w = QWidget()
h = QHBoxLayout(w)
@@ -374,6 +404,25 @@ class MainWindow(QMainWindow):
info.setStyleSheet("color: #a6adc8;")
form.addRow(info)
self.lrc_enabled_cb = QCheckBox("Включён (генерация LRC)")
self.lrc_enabled_cb.setToolTip(
"Отключите, чтобы пропустить этап генерации text.lrc.\n"
"Остальные этапы пайплайна продолжат работать."
)
form.addRow("", self.lrc_enabled_cb)
self.lrc_embed_enabled_cb = QCheckBox("Включён (встраивание LRC в аудио)")
self.lrc_embed_enabled_cb.setToolTip(
"Отключите, чтобы пропустить этап встраивания LRC в теги аудио."
)
form.addRow("", self.lrc_embed_enabled_cb)
self.lrc_skip_existing_cb = QCheckBox("Пропускать готовые text.lrc")
self.lrc_skip_existing_cb.setToolTip(
"Если <output>/<stem>/text.lrc уже существует и содержит корректный LRC,\n"
"файл не будет пересчитан. Папка с metadata.json от другого трека\n"
"готовой не считается."
)
form.addRow("", self.lrc_skip_existing_cb)
self.lrc_audio_source_cb = QComboBox()
self.lrc_audio_source_cb.addItems(["auto", "vocals", "convert", "input"])
self.lrc_audio_source_cb.setToolTip(
@@ -433,6 +482,7 @@ class MainWindow(QMainWindow):
bar.setTabTextColor(2, active if self.convert_enabled_cb.isChecked() else inactive)
bar.setTabTextColor(3, active if self.separate_enabled_cb.isChecked() else inactive)
bar.setTabTextColor(4, active if self.transcribe_enabled_cb.isChecked() else inactive)
bar.setTabTextColor(5, active if self.lrc_enabled_cb.isChecked() else inactive)
def _sync_engine_fields(self) -> None:
is_whisperx = self.engine_cb.currentText() == "whisperx"
@@ -503,6 +553,9 @@ class MainWindow(QMainWindow):
self.lrc_align_model_edit.setText(cfg.lrc.align_model or "")
self.lrc_device_cb.setCurrentText(cfg.lrc.device)
self.lrc_embed_in_place_cb.setChecked(cfg.lrc.embed_in_place)
self.lrc_enabled_cb.setChecked(cfg.lrc.enabled)
self.lrc_embed_enabled_cb.setChecked(cfg.lrc.embed_enabled)
self.lrc_skip_existing_cb.setChecked(cfg.lrc.skip_existing)
self.chunk_size_spin.setValue(cfg.chunk_size)
def _collect_config(self) -> PipelineConfig:
@@ -562,6 +615,9 @@ class MainWindow(QMainWindow):
)
lrc_lang = self.lrc_language_cb.currentText().strip()
cfg.lrc = LrcConfig(
enabled=self.lrc_enabled_cb.isChecked(),
embed_enabled=self.lrc_embed_enabled_cb.isChecked(),
skip_existing=self.lrc_skip_existing_cb.isChecked(),
language=None if lrc_lang in ("", "auto") else lrc_lang,
align_model=self.lrc_align_model_edit.text().strip() or None,
device=self.lrc_device_cb.currentText(),
@@ -642,6 +698,18 @@ class MainWindow(QMainWindow):
"Align-модель доступна только для WhisperX. Переключите движок на whisperx."
)
return
if task == "lrc" and not cfg.lrc.enabled:
QMessageBox.warning(
self, "LRC отключён",
"Этап генерации LRC выключен. Включите его на вкладке LRC."
)
return
if task == "embed" and not cfg.lrc.embed_enabled:
QMessageBox.warning(
self, "Встраивание LRC отключено",
"Этап встраивания LRC выключен. Включите его на вкладке LRC."
)
return
self.log_view.clear()
self._log_device_info()

26
gui/theme.py
View File

@@ -170,29 +170,29 @@ QCheckBox::indicator:checked {
}
QProgressBar {
background-color: #181825;
border: 1px solid #313244;
background-color: #0b0b12;
border: 1px solid #7f849c;
border-radius: 8px;
text-align: center;
color: #cdd6f4;
min-height: 24px;
max-height: 24px;
font-weight: 500;
color: #ffffff;
min-height: 26px;
max-height: 26px;
font-weight: 700;
}
QProgressBar::chunk {
background-color: #89b4fa;
background-color: #1e66f5;
border-radius: 7px;
margin: 1px;
}
QProgressBar#overallProgress {
min-height: 28px;
max-height: 28px;
font-weight: 600;
color: #cdd6f4;
border-color: #45475a;
min-height: 30px;
max-height: 30px;
font-weight: 700;
color: #ffffff;
border: 1px solid #a6adc8;
}
QProgressBar#overallProgress::chunk {
background-color: #a6e3a1;
background-color: #2f9e44;
border-radius: 8px;
margin: 2px;
}