init

auto_subtitle/cli.py

@@ -0,0 +1,101 @@
+import os
+import ffmpeg
+import whisper
+import argparse
+import warnings
+import tempfile
+from .utils import filename, str2bool, write_srt
+
+
+def main():
+    parser = argparse.ArgumentParser(
+        formatter_class=argparse.ArgumentDefaultsHelpFormatter)
+    parser.add_argument("video", nargs="+", type=str,
+                        help="paths to video files to transcribe")
+    parser.add_argument("--model", default="small",
+                        choices=whisper.available_models(), help="name of the Whisper model to use")
+    parser.add_argument("--output_dir", "-o", type=str,
+                        default=".", help="directory to save the outputs")
+    parser.add_argument("--verbose", type=str2bool, default=False,
+                        help="Whether to print out the progress and debug messages")
+
+    parser.add_argument("--task", type=str, default="transcribe", choices=[
+                        "transcribe", "translate"], help="whether to perform X->X speech recognition ('transcribe') or X->English translation ('translate')")
+
+    args = parser.parse_args().__dict__
+    model_name: str = args.pop("model")
+    output_dir: str = args.pop("output_dir")
+    os.makedirs(output_dir, exist_ok=True)
+
+    if model_name.endswith(".en"):
+        warnings.warn(
+            f"{model_name} is an English-only model, forcing English detection.")
+        args["language"] = "en"
+
+    model = whisper.load_model(model_name)
+    audios = get_audio(args.pop("video"))
+    subtitles = get_subtitles(
+        audios, lambda audio_path: model.transcribe(audio_path, **args)
+    )
+    # bash command to download a youtube video with `youtube-dl` and save it as `video.mp4`:
+    # youtube-dl -f 22 -o video.mp4 https://www.youtube.com/watch?v=QH2-TGUlwu4
+
+    for path, srt_path in subtitles.items():
+        out_path = os.path.join(output_dir, f"{filename(path)}.mp4")
+
+        print(f"Adding subtitles to {filename(path)}...")
+
+        video = ffmpeg.input(path)
+        audio = video.audio
+
+        stderr = ffmpeg.concat(
+            video.filter('subtitles', srt_path, force_style="OutlineColour=&H40000000,BorderStyle=3"), audio, v=1, a=1
+        ).output(out_path).run(quiet=True, overwrite_output=True)
+
+        print(f"Saved subtitled video to {os.path.abspath(out_path)}.")
+
+
+def get_audio(paths):
+    temp_dir = tempfile.gettempdir()
+
+    audio_paths = {}
+
+    for path in paths:
+        print(f"Extracting audio from {filename(path)}...")
+        output_path = os.path.join(temp_dir, f"{filename(path)}.wav")
+
+        ffmpeg.input(path).output(
+            output_path,
+            acodec="pcm_s16le", ac=1, ar="16k"
+        ).run(quiet=True, overwrite_output=True)
+
+        audio_paths[path] = output_path
+
+    return audio_paths
+
+
+def get_subtitles(audio_paths: list, transcribe: callable):
+    temp_dir = tempfile.gettempdir()
+    subtitles_path = {}
+
+    for path, audio_path in audio_paths.items():
+        srt_path = os.path.join(temp_dir, f"{filename(path)}.srt")
+
+        print(
+            f"Generating subtitles for {filename(path)}... This might take a while."
+        )
+
+        warnings.filterwarnings("ignore")
+        result = transcribe(audio_path)
+        warnings.filterwarnings("default")
+
+        with open(srt_path, "w", encoding="utf-8") as srt:
+            write_srt(result["segments"], file=srt)
+
+        subtitles_path[path] = srt_path
+
+    return subtitles_path
+
+
+if __name__ == '__main__':
+    main()