feat: one step implement

2026-06-16 13:04:29 +03:30
parent adda6c42b2
commit cad4e681c3
2 changed files with 225 additions and 36 deletions
@@ -1,7 +1,17 @@
 import argparse
+import os
+import platform
+import re
+import shutil
 import subprocess
+import tempfile
 from pathlib import Path

+SRT_BLOCK_RE = re.compile(
+    r"\s*(\d+)\s+([\d:,]+)\s+-->\s+([\d:,]+)\s+(.+?)(?=\n\s*\d+\s+\d\d:|\Z)",
+    re.S,
+)
+

 def resolve_path(value: str) -> Path:
    return Path(value).expanduser().resolve()
@@ -12,39 +22,135 @@ def run(command: list[str]) -> None:
    subprocess.run(command, check=True)


-def main() -> None:
-    parser = argparse.ArgumentParser(
-        description="Replace a video's original audio with a new voiceover audio file."
+def remove_temp_dir(path: Path) -> None:
+    try:
+        shutil.rmtree(path, ignore_errors=True)
+    except PermissionError:
+        pass
+    except OSError:
+        pass
+
+
+def to_seconds(timestamp: str) -> float:
+    h, m, rest = timestamp.split(":")
+    s, ms = rest.split(",")
+    return int(h) * 3600 + int(m) * 60 + int(s) + int(ms) / 1000
+
+
+def read_srt(path: Path) -> list[tuple[float, float, str]]:
+    text = path.read_text(encoding="utf-8-sig")
+    captions = []
+    for _, start, end, body in SRT_BLOCK_RE.findall(text):
+        clean_text = " ".join(line.strip() for line in body.strip().splitlines())
+        captions.append((to_seconds(start), to_seconds(end), clean_text))
+    if not captions:
+        raise ValueError(f"No captions found in {path}")
+    return captions
+
+
+def media_duration(path: Path) -> float:
+    result = subprocess.run(
+        [
+            "ffprobe",
+            "-v", "error",
+            "-show_entries", "format=duration",
+            "-of", "default=noprint_wrappers=1:nokey=1",
+            str(path),
+        ],
+        check=True,
+        capture_output=True,
+        text=True,
    )
-    parser.add_argument("--video", required=True, help="Input video file, for example input/tutorial.mp4")
-    parser.add_argument("--audio", required=True, help="Input voiceover audio file, for example output/voiceover.wav")
-    parser.add_argument("--output", required=True, help="Output MP4 file, for example output/final_video.mp4")
-    parser.add_argument(
-        "--reencode-video",
-        action="store_true",
-        help="Re-encode video with H.264 instead of copying. Use only if the copied output has playback issues.",
-    )
-    args = parser.parse_args()
+    return float(result.stdout.strip())

-    video_file = resolve_path(args.video)
-    audio_file = resolve_path(args.audio)
-    output_file = resolve_path(args.output)

-    if not video_file.exists():
-        raise FileNotFoundError(f"Video not found: {video_file}")
-    if not audio_file.exists():
-        raise FileNotFoundError(f"Audio not found: {audio_file}")
+def ps_quote(value: str) -> str:
+    return "'" + value.replace("'", "''") + "'"

-    output_file.parent.mkdir(parents=True, exist_ok=True)

-    video_codec_args = ["-c:v", "libx264", "-preset", "medium", "-crf", "18"] if args.reencode_video else ["-c:v", "copy"]
+def speak_macos(text: str, voice: str, rate: int, output: Path) -> None:
+    run(["say", "-v", voice, "-r", str(rate), "-o", str(output), text])
+
+
+def speak_windows(text: str, voice: str, rate: int, output: Path) -> None:
+    ps_script = f"""
+Add-Type -AssemblyName System.Speech
+$synth = New-Object System.Speech.Synthesis.SpeechSynthesizer
+$synth.SelectVoice({ps_quote(voice)})
+$synth.Rate = {rate}
+$synth.SetOutputToWaveFile({ps_quote(str(output))})
+$synth.Speak({ps_quote(text)})
+$synth.Dispose()
+"""
+    run(["powershell", "-NoProfile", "-ExecutionPolicy", "Bypass", "-Command", ps_script])
+
+
+def default_voice() -> str:
+    if platform.system() == "Darwin":
+        return "Daniel"
+    if platform.system() == "Windows":
+        return "Microsoft David Desktop"
+    raise RuntimeError("One-step SRT voiceover generation is supported only on macOS and Windows.")
+
+
+def default_rate() -> int:
+    if platform.system() == "Darwin":
+        return 185
+    if platform.system() == "Windows":
+        return 0
+    raise RuntimeError("One-step SRT voiceover generation is supported only on macOS and Windows.")
+
+
+def speak(text: str, voice: str, rate: int, output: Path) -> None:
+    system = platform.system()
+    if system == "Darwin":
+        speak_macos(text, voice, rate, output)
+        return
+    if system == "Windows":
+        speak_windows(text, voice, rate, output)
+        return
+    raise RuntimeError("One-step SRT voiceover generation is supported only on macOS and Windows.")
+
+
+def create_voiceover_from_srt(srt_path: Path, wav_output: Path, voice: str, rate: int, work_dir: Path) -> None:
+    captions = read_srt(srt_path)
+    segment_files = []
+
+    for i, (start, end, text) in enumerate(captions, 1):
+        duration = end - start
+        raw_audio = work_dir / f"raw_{i:03d}.aiff" if platform.system() == "Darwin" else work_dir / f"raw_{i:03d}.wav"
+        segment_wav = work_dir / f"seg_{i:03d}.wav"
+
+        speak(text, voice, rate, raw_audio)
+
+        spoken_duration = media_duration(raw_audio)
+        if spoken_duration > duration:
+            print(f"Warning: caption {i} speech is {spoken_duration:.2f}s but slot is {duration:.2f}s; it will be trimmed.")
+
+        run([
+            "ffmpeg", "-y",
+            "-i", str(raw_audio),
+            "-af", f"apad,atrim=0:{duration}",
+            "-ar", "44100",
+            "-ac", "2",
+            str(segment_wav),
+        ])
+        segment_files.append(segment_wav)
+
+    concat_list = work_dir / "list.txt"
+    concat_list.write_text("".join(f"file '{os.path.abspath(file)}'\n" for file in segment_files), encoding="utf-8")
+    run(["ffmpeg", "-y", "-f", "concat", "-safe", "0", "-i", str(concat_list), "-c:a", "pcm_s16le", str(wav_output)])
+
+
+def combine_video_audio(video_file: Path, audio_file: Path, output_file: Path, reencode_video: bool) -> None:
+    video_codec_args = ["-c:v", "libx264", "-preset", "medium", "-crf", "18"] if reencode_video else ["-c:v", "copy"]

    command = [
        "ffmpeg", "-y",
        "-i", str(video_file),
        "-i", str(audio_file),
-        "-map", "0:v:0",      # video from first input
-        "-map", "1:a:0",      # audio from second input only, so original video audio is muted/removed
+        "-map", "0:v:0",
+        "-map", "1:a:0",
        *video_codec_args,
        "-c:a", "aac",
        "-b:a", "192k",
@@ -52,6 +158,59 @@ def main() -> None:
        str(output_file),
    ]
    run(command)
+
+
+def main() -> None:
+    parser = argparse.ArgumentParser(
+        description="Replace a video's original audio with a voiceover audio file, or generate the voiceover from an SRT first."
+    )
+    parser.add_argument("--video", required=True, help="Input video file, for example input/tutorial.mp4")
+    parser.add_argument("--audio", default=None, help="Existing voiceover audio file, for example output/voiceover.wav")
+    parser.add_argument("--srt", default=None, help="Input SRT file for one-step voiceover generation")
+    parser.add_argument("--output", required=True, help="Output MP4 file, for example output/final_video.mp4")
+    parser.add_argument("--voice", default=None, help="Voice name for one-step SRT mode")
+    parser.add_argument("--rate", type=int, default=None, help="Speech rate for one-step SRT mode")
+    parser.add_argument(
+        "--reencode-video",
+        action="store_true",
+        help="Re-encode video with H.264 instead of copying. Use only if the copied output has playback issues.",
+    )
+    args = parser.parse_args()
+
+    if not args.audio and not args.srt:
+        parser.error("Provide either --audio for the current two-step flow or --srt for one-step generation.")
+    if args.audio and args.srt:
+        parser.error("Use either --audio or --srt, not both.")
+
+    video_file = resolve_path(args.video)
+    output_file = resolve_path(args.output)
+
+    if not video_file.exists():
+        raise FileNotFoundError(f"Video not found: {video_file}")
+
+    output_file.parent.mkdir(parents=True, exist_ok=True)
+
+    if args.audio:
+        audio_file = resolve_path(args.audio)
+        if not audio_file.exists():
+            raise FileNotFoundError(f"Audio not found: {audio_file}")
+        combine_video_audio(video_file, audio_file, output_file, args.reencode_video)
+    else:
+        srt_path = resolve_path(args.srt)
+        if not srt_path.exists():
+            raise FileNotFoundError(f"SRT not found: {srt_path}")
+
+        voice = args.voice or default_voice()
+        rate = args.rate if args.rate is not None else default_rate()
+
+        temp_dir = Path(tempfile.mkdtemp(prefix="video_generator_"))
+        try:
+            temp_wav = temp_dir / "voiceover.wav"
+            create_voiceover_from_srt(srt_path, temp_wav, voice, rate, temp_dir)
+            combine_video_audio(video_file, temp_wav, output_file, args.reencode_video)
+        finally:
+            remove_temp_dir(temp_dir)
+
    print(f"Done: {output_file}")