FastDub 3.3.0

A Python CLI package for voice over subtitles, with the ability to embed in video, audio ducking, and dynamic voice changer for a single track; auto translating; download and upload to YouTube supports

Package for voice over subtitles:

• with the ability to embed in video,
• audio ducking,
• dynamic voice changer for a single track. Add --voice-set-anchor at the beginning of the subtitle line. (Applies to all subsequent ones)

pip install -U FastDub

---

Install for Ubuntu

sudo apt update -y && sudo apt upgrade -y

sudo apt install python3 python3-pip ffmpeg espeak libxml2-dev libxslt1-dev

; libxml for translator functions

sudo python3 -m pip install -U pip setuptools wheel

sudo python3 -m pip install -U FastDub

Install all dependencies

pip install FastDub[ALL] # default

Youtube support

Youtube argument group

pip install FastDub[YT]

Process all videos from a channel

To get the channel id if it only has a username, run the JS code in the browser console:

document.querySelector("body>link").href

Then, in the input parameter, you can pass a link to the channel (with an id, not with a username) or a string of the format "c/CHANNEL_ID"

For example:

python -m fastdub -i "c/UCi-5OZ2tYuwMLIcEyOsbdRA" -yt

YouTube Search argument group

That the input data was a request to YouTube, they must begin with "?".

For Example:

python -m fastdub -i "?#python" -yt -yts-l 5

YouTube Upload argument group

pip install FastDub[YTU]

To download, you need to go to credentials (Create a new project if needed) > + Create credentials > OAuth Client ID > Desktop App

After filling in the required fields, DOWNLOAD JSON format and save to a working folder under the name client_secrets.json

• Specifying the -ytu-ps argument as non-private may result in processing errors, but this was not observed during testing.

• Please note that video uploads require a large amount of quota ( default has 10,000 per day)

Subtitles translate

Translate argument group

pip install FastDub[TR]

Usage

python -m fastdub --help

usage: fastdub [-h] [-rc {0,1,2}] [-ra | --cleanup-audio | -n-ra | --no-cleanup-audio] [-l LANGUAGE]
               [-tc THREADS_COUNT] -i INPUT [-vf VIDEO_FORMAT] [-sf SUBTITLES_FORMAT] [-En EXCLUDE [EXCLUDE ...]]
               [-Eu EXCLUDE_UNDERSCORE] [-sc | --side-chain | -n-sc | --no-side-chain] [-sc-msl MIN_SILENCE_LEN]
               [-sc-st SILENCE_THRESH] [-sc-gdo GAIN_DURING_OVERLAY]
               [-v {microsoft irina desktop - russian,microsoft zira desktop - english united states),microsoft david desktop - english (united states,aleksandr-hq,arina,artemiy,evgeniy-eng,evgeniy-rus,lyubov,marianna,mikhail,pavel,tatiana,victoria,vitaliy,volodymyr,yuriy}]
               [-a ALIGN] [-v-set-a VOICE_SET_ANCHOR] [-ll {trace,debug,verbose,info,warning,error,fatal,panic,quiet}]
               [-y | --confirm | -n-y | --no-confirm] [-af AUDIO_FORMAT] [-tb | --traceback | -n-tb | --no-traceback]
               [-yt] [-ak API_KEYS [API_KEYS ...]] [-yts] [-yts-l YOUTUBE_SEARCH_LIMIT]
               [-yts-rg YOUTUBE_SEARCH_REGION] [-ytu] [-ytu-ps {private,public,unlisted}] [-ytu-t] [-tr]
               [--rewrite-srt | --no-rewrite-srt]
               [-ts {alibaba,argos,baidu,bing,caiyun,deepl,google,iciba,iflytek,itranslate,lingvanex,mglip,niutrans,papago,reverso,sogou,tencent,translateCom,utibet,yandex,youdao}]

fastdub is a tool for dubbing videos by subtitle files.

options:
  -h, --help            show this help message and exit
  -rc {0,1,2}, --remove-cache {0,1,2}
                        Remove all cache (_cached_texts directory) files
                                0 - No remove cache
                                1 - Delete cache before voice acting
                                2 - Delete cache after voice acting (default)
  -ra, --cleanup-audio, -n-ra, --no-cleanup-audio
                        Remove result audio if video exists (default: False)
  -l LANGUAGE, --language LANGUAGE
                        Subtitles language (ru)
  -tc THREADS_COUNT, --threads-count THREADS_COUNT
                        Process count to download (pass to cpu count, < 2 to disable)
                                *N = N * cpu count

Input:
  -i INPUT, --input INPUT
                        Input directory/YouTube Playlist/Video URL.
  -vf VIDEO_FORMAT, --video-format VIDEO_FORMAT
                        Video format (default: .mp4).
  -sf SUBTITLES_FORMAT, --subtitles-format SUBTITLES_FORMAT
                        Subtitles format (default: .srt).

Audio Ducking:
  -sc, --side-chain, -n-sc, --no-side-chain
                        Enable audio side chain compress (ducking) (default: True)
  -sc-msl MIN_SILENCE_LEN, --min-silence-len MIN_SILENCE_LEN, --attack MIN_SILENCE_LEN
                        Minimum silence length in ms (default: 100)
  -sc-st SILENCE_THRESH, --silence-thresh SILENCE_THRESH
                        Silence threshold in dB
  -sc-gdo GAIN_DURING_OVERLAY, --gain-during-overlay GAIN_DURING_OVERLAY
                        Gain during overlay in dB (default: -11)

Voicer:
  -v {microsoft irina desktop - russian,microsoft zira desktop - english (united states),microsoft david desktop - english (united states),aleksandr-hq,arina,artemiy,evgeniy-eng,evgeniy-rus,lyubov,marianna,mikhail,pavel,tatiana,victoria,vitaliy,volodymyr,yuriy}, --voice {microsoft irina desktop - russian,microsoft zira desktop - english (united states),microsoft david desktop - english (united states),aleksandr-hq,arina,artemiy,evgeniy-eng,evgeniy-rus,lyubov,marianna,mikhail,pavel,tatiana,victoria,vitaliy,volodymyr,yuriy}
                        SAPI voice for voice acting.
  -a ALIGN, --align ALIGN
                        Audio fit align (divisor)
                                1 = right
                                2 = center (default)
  -v-set-a VOICE_SET_ANCHOR, --voice-set-anchor VOICE_SET_ANCHOR
                        Anchor indicating voice actor change (default "!:")

FFMpeg Output:
  -ll {trace,debug,verbose,info,warning,error,fatal,panic,quiet}, --loglevel {trace,debug,verbose,info,warning,error,fatal,panic,quiet}
                        FFMpegWrapper loglevel
  -y, --confirm, -n-y, --no-confirm
                        Don't ask for confirmation (default: True)
  -af AUDIO_FORMAT, --audio-format AUDIO_FORMAT
                        Out audio files format (default wav)

Terminal Output:
  -tb, --traceback, -n-tb, --no-traceback
                        Show debug traceback (default: False)

YouTube:
  -yt, --youtube
  -ak API_KEYS [API_KEYS ...], --api-keys API_KEYS [API_KEYS ...]
                        Youtube API key/s

YouTube Search:
  -yts, --youtube-search
                        The input data is a query for searching on YouTube. (Adds "?" at the start of input)
  -yts-l YOUTUBE_SEARCH_LIMIT, --youtube-search-limit YOUTUBE_SEARCH_LIMIT
                        Sets limit to the number of results. Defaults to 20.
  -yts-rg YOUTUBE_SEARCH_REGION, --youtube-search-region YOUTUBE_SEARCH_REGION
                        Sets the result region. Defaults to "US".

YouTube Upload:
  -ytu, --youtube-upload
                        yt_upload video to YouTube channel after voice acting.
  -ytu-ps {private,public,unlisted}, --privacy-status {private,public,unlisted}
                        Video privacy status (If not private, errors are possible)
  -ytu-t, --youtube-upload-translate
                        Translate title and description on upload. (+ Arguments from translate argument group)

Translate:
  -tr, --translate      Translate input subtitles files
  --rewrite-srt, --no-rewrite-srt
                        Rewrite input subtitles files.
                        If not, add "_" to the beginning of the original subtitle file. (default: False)
  -ts {alibaba,argos,baidu,bing,caiyun,deepl,google,iciba,iflytek,itranslate,lingvanex,mglip,niutrans,papago,reverso,sogou,tencent,translateCom,utibet,yandex,youdao}, --translate-service {alibaba,argos,baidu,bing,caiyun,deepl,google,iciba,iflytek,itranslate,lingvanex,mglip,niutrans,papago,reverso,sogou,tencent,translateCom,utibet,yandex,youdao}
                        Subtitle translation service. (default google)

If the voice set after "!:" is not selected during voiceover, clear the cache with the -rc argument

Example

python -m fastdub -i DirToDub -v "Yuriy" --no-sidechain -sf vtt -vf mkv

All files in the folder will be voices (*.mkv, *.vtt)

Default is mp4 video and srt subtitles.

Then: name and name2 will be announced, and the results will be saved to a folder with the appropriate name.

YT Example

python -m fastdub -yt -i https://youtu.be/DD5UKQggXTc -v "Yuriy" -l ru

YouTube Search Example

python -m fastdub -yt -yts -i "#annoyingorange" -l ru

Translate Example

python -m fastdub -i DirToDub -tr -ts iciba -l ru

Default service is Google.

Other

The fastdub.utils module stores helper functions.

python -m pydoc -w fastdub

You can write your issues on GitHub in English or in Russian.

CHANGELOG

---

3.3.0

General

• Реализован класс GlobalSettings

Audio Processing

• Функция calc_speed_change_ffmpeg_arg по расчету аргумента -af ffmpeg для изменения скорости изменена.

  Значительно ускорена с применением другого алгоритма.

• Оптимизирована функция side_chain.

Voiceover Process

• Подгон длины аудио и TTS теперь объеденины в один процесс для увеличения производительности.
• Теперь создаётся подпапка _result/_working_dir в которой сохраняются отдельные уже подогнанные строки субтитров. После чего все части склеиваются ffmpeg-ом. Значительно уменьшено потребление памяти.
• moviepy больше не является зависимость, вместо этого парсится вывод ffmpeg

FFMpeg Wrapper

• Теперь все аргументы преобразуются в строки перед вызовом check_call

3.2.0

CLI

• Опущены необязательные type=str аргументы.
• Изменена логика BooleanOptionalAction, теперь доступно как --no- так и -n- отрицание.
• Аргумент --cleanup-audio теперь BooleanOptionalAction (по умолчанию True), --cleanup-level удалён.
• Аргумент --voice-set-anchor (по умолчанию так же !:).
• Аргумент --audio-format для выходного аудио (по умолчанию wav).
• Аргумент --sidechain переименован в --side-chain.
• Добавлено сокращение аргументу --traceback - -tb.

Voiceover Process

• Bugfix.
• Оптимизация подгона длительности дорожек (Fitting-а).

TTS

• anchor аргумент (см. --voice-set-anchor).
• Оптимизация смены голосов.

Submodules

YouTube

• При multi-загрузке полностью загруженные видео удаляются с экрана (чтоб не занимать зря место).