Upload lyric_sync/cli.py

lyric_sync/cli.py

@@ -0,0 +1,198 @@
+"""
+Command-line interface for lyric_sync.
+"""
+
+import argparse
+import logging
+import os
+import sys
+
+from lyric_sync.pipeline import LyricSyncPipeline
+
+
+def main():
+    parser = argparse.ArgumentParser(
+        prog="lyric-sync",
+        description="Automatic perfect song lyric acquisition and synchronization.",
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+        epilog="""
+Examples:
+  # Full automatic pipeline (requires AcoustID key + GPU)
+  lyric-sync song.mp3 --acoustid-key YOUR_KEY --output synced.lrc
+
+  # With known metadata (skip identification)
+  lyric-sync song.mp3 --artist "Queen" --title "Bohemian Rhapsody" -o synced.lrc
+
+  # CPU-only, simpler transcription
+  lyric-sync song.mp3 --device cpu --backend whisper --artist "Artist" --title "Title"
+
+  # JSON output for programmatic use
+  lyric-sync song.mp3 --format json --artist "Artist" --title "Title" -o synced.json
+
+  # ASS karaoke subtitles
+  lyric-sync song.mp3 --format ass --artist "Artist" --title "Title" -o karaoke.ass
+""",
+    )
+
+    parser.add_argument("audio", help="Path to audio file (mp3, flac, wav, etc.)")
+    parser.add_argument("-o", "--output", help="Output file path (default: stdout)")
+    parser.add_argument(
+        "--format",
+        choices=["lrc", "lrc_standard", "json", "srt", "ass"],
+        default="lrc",
+        help="Output format (default: lrc = enhanced word-level LRC)",
+    )
+
+    # Metadata overrides
+    meta = parser.add_argument_group("song metadata (skip identification)")
+    meta.add_argument("--artist", help="Artist name")
+    meta.add_argument("--title", help="Track title")
+    meta.add_argument("--album", help="Album name")
+
+    # API keys
+    keys = parser.add_argument_group("API keys")
+    keys.add_argument(
+        "--acoustid-key",
+        default=os.environ.get("ACOUSTID_API_KEY"),
+        help="AcoustID API key (or set ACOUSTID_API_KEY env var)",
+    )
+    keys.add_argument(
+        "--genius-token",
+        default=os.environ.get("GENIUS_TOKEN"),
+        help="Genius API token (or set GENIUS_TOKEN env var)",
+    )
+
+    # Model configuration
+    model = parser.add_argument_group("model configuration")
+    model.add_argument(
+        "--backend",
+        choices=["whisperx", "whisper", "granite"],
+        default="whisperx",
+        help="Transcription backend (default: whisperx)",
+    )
+    model.add_argument(
+        "--whisper-model",
+        default="large-v2",
+        help="Whisper model size (default: large-v2, best for lyrics)",
+    )
+    model.add_argument(
+        "--demucs-model",
+        default="htdemucs_ft",
+        help="Demucs model (default: htdemucs_ft, best quality)",
+    )
+    model.add_argument(
+        "--device",
+        default="cuda",
+        help="Device: cuda or cpu (default: cuda)",
+    )
+    model.add_argument(
+        "--language",
+        default="en",
+        help="Language code (default: en)",
+    )
+
+    # Processing options
+    proc = parser.add_argument_group("processing options")
+    proc.add_argument(
+        "--no-refine",
+        action="store_true",
+        help="Skip onset/offset timing refinement",
+    )
+    proc.add_argument(
+        "--line-break-gap",
+        type=float,
+        default=1.0,
+        help="Seconds of silence to trigger line break in output (default: 1.0)",
+    )
+
+    # Verbosity
+    parser.add_argument(
+        "-v", "--verbose",
+        action="count",
+        default=0,
+        help="Increase verbosity (-v info, -vv debug)",
+    )
+    parser.add_argument(
+        "-q", "--quiet",
+        action="store_true",
+        help="Suppress all logging",
+    )
+
+    args = parser.parse_args()
+
+    # Configure logging
+    if args.quiet:
+        log_level = logging.CRITICAL
+    elif args.verbose >= 2:
+        log_level = logging.DEBUG
+    elif args.verbose >= 1:
+        log_level = logging.INFO
+    else:
+        log_level = logging.WARNING
+
+    logging.basicConfig(
+        level=log_level,
+        format="%(asctime)s [%(levelname)s] %(name)s: %(message)s",
+        datefmt="%H:%M:%S",
+    )
+
+    # Validate inputs
+    if not os.path.isfile(args.audio):
+        print(f"Error: Audio file not found: {args.audio}", file=sys.stderr)
+        sys.exit(1)
+
+    if not args.artist and not args.title and not args.acoustid_key:
+        print(
+            "Warning: No --artist/--title and no --acoustid-key provided. "
+            "Song identification may fail. Will use raw transcription as fallback.",
+            file=sys.stderr,
+        )
+
+    # Run pipeline
+    pipeline = LyricSyncPipeline(
+        acoustid_key=args.acoustid_key,
+        genius_token=args.genius_token,
+        transcription_backend=args.backend,
+        whisper_model=args.whisper_model,
+        demucs_model=args.demucs_model,
+        device=args.device,
+        refine_timings_enabled=not args.no_refine,
+        language=args.language,
+    )
+
+    result = pipeline.sync(
+        audio_path=args.audio,
+        artist=args.artist,
+        title=args.title,
+        album=args.album,
+        output_format=args.format,
+        output_path=args.output,
+    )
+
+    # Print to stdout if no output file specified
+    if not args.output:
+        formatters = {
+            "lrc": lambda: result.to_lrc(enhanced=True, line_break_gap=args.line_break_gap),
+            "lrc_standard": lambda: result.to_lrc(enhanced=False, line_break_gap=args.line_break_gap),
+            "json": lambda: result.to_json(),
+            "srt": lambda: result.to_srt(line_break_gap=args.line_break_gap),
+            "ass": lambda: result.to_ass(line_break_gap=args.line_break_gap),
+        }
+        print(formatters[args.format]())
+
+    # Print summary to stderr
+    if not args.quiet:
+        print(f"\n--- Sync Summary ---", file=sys.stderr)
+        if result.song:
+            print(f"Song: {result.song.artist} - {result.song.title}", file=sys.stderr)
+            print(f"Identified via: {result.song.method}", file=sys.stderr)
+        print(f"Words: {len(result.synced_words)}", file=sys.stderr)
+        print(f"Quality: {result.quality_score:.2f}", file=sys.stderr)
+        if result.alignment_stats:
+            print(f"Direct matches: {result.alignment_stats.match_rate:.1%}", file=sys.stderr)
+            print(f"Coverage: {result.alignment_stats.coverage:.1%}", file=sys.stderr)
+        print(f"Time: {result.processing_time_sec:.1f}s", file=sys.stderr)
+
+
+if __name__ == "__main__":
+    main()