Subpipe/config.json at main · Ceryunus/Subpipe · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
{
  "paths": {
    "video_dir": "data/video",
    "video_with_subs_dir": "data/video_with_subs",
    "audio_dir": "data/audio",
    "subs_original_dir": "data/subs_original",
    "subs_translated_dir": "data/subs_translated",
    "logs_dir": "data/logs",
    "video_file": "sample.mp4",
    "audio_file": "sample.mp3",
    "subs_translated_file_srt": "translated.srt",
    "subs_translated_file_txt": "translated.txt",
    "subs_original_file": "original.srt",
    "subs_original_file_json": "segments.json",
    "output_soft": "sample_output_soft.mp4",
    "output_burned": "sample_output_burned.mp4"
  },
  "ffmpeg": {
    "_comment_ffmpeg": "FFmpeg settings for subtitle rendering. 'crf' controls video quality (0-51, lower is better). 'preset' affects encoding speed (e.g., ultrafast, fast, medium, slow).",
    "font_name": "Arial",
    "font_size": 24,
    "primary_colour": "&H00FFFFFF",
    "outline_colour": "&H00000000",
    "border_style": 3,
    "outline": 1,
    "shadow": 0,
    "preset": "medium",
    "crf": 23,
    "pix_fmt": "yuv420p",
    "subtitle_position": "bottom",
    "_comment_subtitle_position": "Position of burned subtitles: 'bottom', 'top', or 'center'.",
    "hwaccel": "auto",
    "_comment_hwaccel": "Hardware acceleration for FFmpeg: 'auto', 'cuda', or 'none'."
  },
  "models": {
    "_comment_source_lang": "Source language for transcription. Set to 'auto' for auto-detection or use 'en','jp' (e.g., 'de'). See https://huggingface.co/openai/whisper-large-v3 for 99 supported languages.",
    "source_lang": "auto",

    "_comment_whisper_model": "Whisper model for transcription: tiny, base, small, medium, large-v3, large-v3-turbo. Larger models are more accurate but slower.",
    "whisper_model": "medium",

    "_comment_translation_model": "Translation model: 'facebook/nllb-200-1.3B' (accurate) or 'facebook/nllb-200-distilled-600M' (faster). Supports 196 languages, see https://huggingface.co/facebook/nllb-200-1.3B.",
    "translation_model": "facebook/nllb-200-1.3B",

    "_comment_target_lang": "Target language for translation in NLLB format (e.g., 'tur_Latn', 'eng_Latn', 'jpn_Jpan'). Do NOT use short codes like 'tr', 'en'.",
    "target_lang": "tur_Latn",

    "language_code_map": {
      "_comment_language_code_map": "Maps short language codes to NLLB format. Add more languages as needed.",
      "en": "eng_Latn",
      "tr": "tur_Latn",
      "es": "spa_Latn",
      "fr": "fra_Latn",
      "de": "deu_Latn",
      "ja": "jpn_Jpan",
      "zh": "zho_Hans",
      "ko": "kor_Hang",
      "ar": "ara_Arab",
      "ru": "rus_Cyrl"
    }
  },
  "logging": {
    "level": "DEBUG",
    "format": "%(asctime)s - %(levelname)s - %(message)s",
    "max_file_size": "10MB",
    "log_file": "pipeline.log",
    "_comment_log_file": "Name of the unified log file for all pipeline steps."
  },
  "subtitles": {
    "_comment_mode": "Subtitle type: 'soft' (embedded, toggleable) or 'burned' (hardcoded into video).",
    "mode": "burned",
    "_comment_max_line_length": "Maximum characters per subtitle line for readability.",
    "max_line_length": 60,
    "_comment_max_lines": "Maximum number of subtitle lines displayed at once.",
    "max_lines": 2
  },
  "audio_extraction": {
    "use_ffmpeg": true,
    "_comment_use_ffmpeg": "Use FFmpeg instead of MoviePy for audio extraction (faster with hwaccel)."
  },
  "whisper_transcription": {
    "_comment_whisper_transcription": "Parameters for Whisper transcription model.",

    "_comment_beam_size": "Number of beams for beam search. Higher values increase accuracy but slow down transcription.",
    "beam_size": 5,

    "_comment_vad_filter": "Enable Voice Activity Detection to filter silence.",
    "vad_filter": true,

    "_comment_word_timestamps": "Include word-level timestamps in transcription output.",
    "word_timestamps": true,

    "_comment_min_silence_duration_ms": "Minimum silence duration (ms) for VAD filter.",
    "min_silence_duration_ms": 500,

    "_comment_max_initial_timestamp": "Maximum initial timestamp (seconds) for transcription start.",
    "max_initial_timestamp": 30.0

  }
}