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

add resume support, metadata paths, LRCLIB order and overall progress

Browse Source 
- store source track path in metadata.json and validate it when skipping ready files
- add skip_existing/resume_partial options to config and GUI
- identify lyrics via embedded text, Shazam, then one LRCLIB author search
- keep post-ASR LRCLIB text matching and allow cancelling safely
- recover from CUDA OOM and free VRAM between batch stages
- add overall completed/total file progress bar to GUI and pipelines
This commit is contained in:
jadssss
2026-10-25 20:24:01 +03:00
parent 199d12a274
commit 15c0ba5d66
8 changed files with 765 additions and 225 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
core/config.py
View File

@@ -81,6 +81,9 @@ class PipelineConfig:
keep_intermediate: bool = True
overwrite: bool = True
mode: str = "sequential"
chunk_size: int = 64
skip_existing: bool = False # skip files that already have text.txt
resume_partial: bool = True # reuse existing vocals.wav / convert.wav
convert: ConvertConfig = field(default_factory=ConvertConfig)
separate: SeparateConfig = field(default_factory=SeparateConfig)

21
core/lrc_pipeline.py
View File

@@ -16,7 +16,8 @@ from .transcribe import transcribe_aligned
LogFn = Callable[[str], None]
ProgressFn = Callable[[int, int, str], None]
ProgressFn = Callable[[str, int, int], None] # stage, current, total
OverallFn = Callable[[int, int], None] # completed_files, total_files
def _pick_audio_source(
@@ -49,6 +50,7 @@ def run_lrc_pipeline(
log: LogFn = print,
on_progress: Optional[ProgressFn] = None,
stop_event: Optional[threading.Event] = None,
on_overall_progress: Optional[OverallFn] = None,
) -> list[FileResult]:
input_dir = Path(config.input_dir)
if not input_dir.is_dir():
@@ -67,13 +69,15 @@ def run_lrc_pipeline(
results: list[FileResult] = []
total = len(files)
if on_overall_progress:
on_overall_progress(0, total)
for idx, f in enumerate(files, 1):
if stop_event is not None and stop_event.is_set():
log("⏹ Отменено")
break
if on_progress:
on_progress(idx - 1, total, "lrc")
on_progress("lrc", idx - 1, total)
log(f"\n▶ [{idx}/{total}] {f.name}")
track_out = output_dir / f.stem
@@ -127,7 +131,9 @@ def run_lrc_pipeline(
finally:
shutil.rmtree(tmp, ignore_errors=True)
if on_progress:
on_progress(idx, total, "lrc")
on_progress("lrc", idx, total)
if on_overall_progress:
on_overall_progress(idx, total)
return results
@@ -137,6 +143,7 @@ def run_embed_pipeline(
log: LogFn = print,
on_progress: Optional[ProgressFn] = None,
stop_event: Optional[threading.Event] = None,
on_overall_progress: Optional[OverallFn] = None,
) -> list[FileResult]:
input_dir = Path(config.input_dir)
if not input_dir.is_dir():
@@ -154,13 +161,15 @@ def run_embed_pipeline(
results: list[FileResult] = []
total = len(files)
if on_overall_progress:
on_overall_progress(0, total)
for idx, f in enumerate(files, 1):
if stop_event is not None and stop_event.is_set():
log("⏹ Отменено")
break
if on_progress:
on_progress(idx - 1, total, "embed")
on_progress("embed", idx - 1, total)
lrc_path = output_dir / f.stem / "text.lrc"
error: Optional[str] = None
@@ -190,6 +199,8 @@ def run_embed_pipeline(
results.append(FileResult(f, None, None, error, "embed"))
finally:
if on_progress:
on_progress(idx, total, "embed")
on_progress("embed", idx, total)
if on_overall_progress:
on_overall_progress(idx, total)
return results

792
core/pipeline.py
View File

File diff suppressed because it is too large Load Diff

5
core/separate.py
View File

@@ -91,13 +91,10 @@ def separate_vocals_batch(
total = len(input_wavs)
processed = 0
while proc.poll() is None:
time.sleep(2.0)
if progress_callback and processed < total - 1:
if proc.wait(2.0) is not None:
break
processed += 1
progress_callback(processed)
else:
proc.wait(2.0)
if proc.returncode != 0:
raise RuntimeError(f"unblend batch failed with exit code {proc.returncode}")

52
core/transcribe.py
View File

@@ -11,6 +11,31 @@ from pathlib import Path
from typing import Optional
# Set permanently once the CUDA context has died (e.g. after an unrecoverable
# OOM). Every later CUDA transcription fails instantly instead of re-loading
# the model forever and hanging the pipeline.
_CUDA_BROKEN = False
def _is_cuda_error(exc: BaseException) -> bool:
err = str(exc).lower()
return (
"out of memory" in err
or "cuda" in err
or "device ordinal" in err
or "cublas" in err
or "cudnn" in err
)
def _check_cuda_usable(device: str) -> None:
if device == "cuda" and _CUDA_BROKEN:
raise RuntimeError(
"CUDA context is broken (previous OOM was unrecoverable). "
"Restart the app or switch device to CPU."
)
def _resolve_device(device: str) -> str:
if device in ("cuda", "cpu"):
return device
@@ -39,7 +64,7 @@ def _free_vram() -> None:
import torch
if torch.cuda.is_available():
torch.cuda.empty_cache()
except ImportError:
except Exception: # noqa: BLE001 — cleanup must never mask the original error
pass
@@ -124,7 +149,9 @@ def transcribe(
local_files_only: bool = False,
vad_filter: bool = True,
) -> str:
global _CUDA_BROKEN
device = _resolve_device(device)
_check_cuda_usable(device)
if engine == "whisperx":
try:
@@ -133,16 +160,26 @@ def transcribe(
batch_size, local_files_only, vad_filter,
)
except Exception as e:
err = str(e).lower()
if "out of memory" in err or "cuda" in err or "device ordinal" in err:
print(f" [ASR] CUDA error: {e}. Clearing model cache and retrying...")
_WhisperXModels._cache.clear()
_free_vram()
if not _is_cuda_error(e):
raise
print(f" [ASR] CUDA error: {e}. Clearing model cache and retrying...")
_WhisperXModels._cache.clear()
_free_vram()
try:
return _transcribe_whisperx(
audio_path, model, language, device, compute_type,
batch_size, local_files_only, vad_filter,
)
raise
except Exception as retry_error:
if not _is_cuda_error(retry_error):
raise
_CUDA_BROKEN = True
print(" [ASR] CUDA context is unrecoverable — "
"GPU disabled for the rest of this run.")
raise RuntimeError(
"CUDA context is broken after OOM recovery failed. "
"Restart the app or switch device to CPU."
) from retry_error
if engine == "gigaam":
return _transcribe_gigaam(audio_path, gigaam_provider)
@@ -191,6 +228,7 @@ def transcribe_aligned(
"""
import whisperx
device = _resolve_device(device)
_check_cuda_usable(device)
model_obj = _WhisperXModels.get(model, device, compute_type, local_files_only)
audio = whisperx.load_audio(str(audio_path))
result = model_obj.transcribe(audio, batch_size=batch_size, language=language)