subliminal - Man Page

Examples (TL;DR)


subliminal — subliminal Documentation

Subliminal is a python 2.7+ library to search and download subtitles. It comes with an easy to use yet powerful CLI suitable for direct use or cron jobs.




Download English subtitles:

$ subliminal download -l en The.Big.Bang.Theory.S05E18.HDTV.x264-LOL.mp4
Collecting videos  [####################################]  100%
1 video collected / 0 video ignored
Downloading subtitles  [####################################]  100%
Downloaded 1 subtitle

For cron usage, make sure to specify a maximum age (with --age) so subtitles are searched for recent videos only. Otherwise you will get banned from the providers for abuse due to too many requests. If subliminal didn’t find subtitles for an old video, it’s unlikely it will find subtitles for that video ever anyway.

See CLI for more details on the available commands and options.

Nautilus/Nemo integration

See the dedicated project page for more information.

High level API

You can call subliminal in many different ways depending on how much control you want over the process. For most use cases, you can stick to the standard API.


Let’s start by importing subliminal:

>>> import os
>>> from babelfish import *
>>> from subliminal import *

Before going further, there are a few things to know about subliminal.


The Movie and Episode classes represent a video, existing or not. You can create a video by name (or path) with Video.fromname, use scan_video() on an existing file path to get even more information about the video or use scan_videos() on an existing directory path to scan a whole directory for videos.

>>> video = Video.fromname('The.Big.Bang.Theory.S05E18.HDTV.x264-LOL.mp4')
>>> video
<Episode [The Big Bang Theory s05e18]>

Here video information was guessed based on the name of the video, you can access some video attributes:

>>> video.video_codec
>>> video.release_group


Before proceeding to listing and downloading subtitles, you need to configure the cache. Subliminal uses a cache to reduce repeated queries to providers and improve overall performance with no impact on search quality. For the sake of this example, we’re going to use a memory backend.

>>> my_region = region.configure('dogpile.cache.memory')

Choose a cache that fits your application and prefer persistent over volatile backends. The file backend is usually a good choice. See dogpile.cache’s documentation for more details on backends.

Now that we’re done with the basics, let’s have some real fun.


To list subtitles, subliminal provides a list_subtitles() function that will return all found subtitles:

>>> subtitles = list_subtitles([video], {Language('hun')}, providers=['podnapisi'])
>>> subtitles[video]
[<PodnapisiSubtitle 'ZtAW' [hu]>, <PodnapisiSubtitle 'ONAW' [hu]>]

As you noticed, all parameters are iterables but only contain one item which means you can deal with a lot of videos, languages and providers at the same time. For the sake of this example, we filter providers to use only one, pass providers=None (default) to search on all providers.


It’s usual you have multiple candidates for subtitles. To help you chose which one to download, subliminal can compare them to the video and tell you exactly what matches with get_matches():

>>> for s in subtitles[video]:
...     sorted(s.get_matches(video))
['country', 'episode', 'release_group', 'season', 'series', 'source', 'video_codec', 'year']
['country', 'episode', 'season', 'series', 'source', 'year']

And then compute a score with those matches with compute_score():

>>> for s in subtitles[video]:
...     {s: compute_score(s, video)}
{<PodnapisiSubtitle 'ZtAW' [hu]>: 789}
{<PodnapisiSubtitle 'ONAW' [hu]>: 772}

Now you should have a better idea about which one you should choose.


We can settle on the first subtitle and download its content using download_subtitles():

>>> subtitle = subtitles[video][0]
>>> subtitle.content is None
>>> download_subtitles([subtitle])
>>> subtitle.content.split(b'\n')[2]
b'Elszaladok a boltba'

If you want a string instead of bytes, you can access decoded content with the text property:

>>> subtitle.text.split('\n')[3]
'néhány apróságért.'

Downloading best subtitles

Downloading best subtitles is what you want to do in almost all cases, as a shortcut for listing, scoring and downloading you can use download_best_subtitles():

>>> best_subtitles = download_best_subtitles([video], {Language('hun')}, providers=['podnapisi'])
>>> best_subtitles[video]
[<PodnapisiSubtitle 'ZtAW' [hu]>]
>>> best_subtitle = best_subtitles[video][0]
>>> best_subtitle.content.split(b'\n')[2]
b'Elszaladok a boltba'

We end up with the same subtitle but with one line of code. Neat.


We got ourselves a nice subtitle, now we can save it on the file system using save_subtitles():

>>> save_subtitles(video, [best_subtitle])
[<PodnapisiSubtitle 'ZtAW' [hu]>]
>>> os.listdir()

How it works


Subliminal uses multiple providers to give users a vast choice and have a better chance to find the best matching subtitles. Current supported providers are:

  • Addic7ed
  • LegendasTV
  • NapiProjekt
  • OpenSubtitles
  • Podnapisi
  • Shooter
  • TheSubDB
  • TvSubtitles

Providers all inherit the same Provider base class and thus share the same API. They are registered on the subliminal.providers entry point and are exposed through the provider_manager for easy access.

To work with multiple providers seamlessly, the ProviderPool exposes the same API but distributes it to its providers and AsyncProviderPool does it asynchronously.


Rating subtitles and comparing them is probably the most difficult part and this is where subliminal excels with its powerful scoring algorithm.

Using guessit and enzyme, subliminal extracts properties of the video and match them with the properties of the subtitles found with the providers.

Equations in subliminal.score give a score to each property (called a match). The more matches the video and the subtitle have, the higher the score computed with compute_score() gets.


Various libraries are used by subliminal and are key to its success:

  • guessit to guess information from filenames
  • enzyme to detect embedded subtitles in videos and read other video metadata
  • babelfish to work with languages
  • requests to make human readable HTTP requests
  • BeautifulSoup to parse HTML and XML
  • dogpile.cache to cache intermediate search results
  • stevedore to manage the provider entry point
  • chardet to detect subtitles’ encoding
  • pysrt to validate downloaded subtitles



Usage: subliminal [OPTIONS] COMMAND [ARGS]...

  Subtitles, faster than your thoughts.

  --addic7ed USERNAME PASSWORD    Addic7ed configuration.
  --legendastv USERNAME PASSWORD  LegendasTV configuration.
  --opensubtitles USERNAME PASSWORD
                                  OpenSubtitles configuration.
  --omdb APIKEY                   OMDB API key.
  --cache-dir DIRECTORY           Path to the cache directory.  [default:
  --debug                         Print useful information for debugging
                                  subliminal and for reporting bugs.
  --version                       Show the version and exit.
  --help                          Show this message and exit.

  cache     Cache management.
  download  Download best subtitles.

  Suggestions and bug reports are greatly appreciated:

subliminal download

Usage: subliminal download [OPTIONS] PATH...

  Download best subtitles.

  PATH can be an directory containing videos, a video file path or a video
  file name. It can be used multiple times.

  If an existing subtitle is detected (external or embedded) in the correct
  language, the download is skipped for the associated video.

  -l, --language LANGUAGE         Language as IETF code, e.g. en, pt-BR (can
                                  be used multiple times).  [required]
  -p, --provider [addic7ed|argenteam|legendastv|opensubtitles|opensubtitlesvip|podnapisi|shooter|thesubdb|tvsubtitles]
                                  Provider to use (can be used multiple
  -r, --refiner [hash|metadata|omdb|tvdb]
                                  Refiner to use (can be used multiple times).
  -a, --age AGE                   Filter videos newer than AGE, e.g. 12h,
  -d, --directory DIR             Directory where to save subtitles, default
                                  is next to the video file.
  -e, --encoding ENC              Subtitle file encoding, default is to
                                  preserve original encoding.
  -s, --single                    Save subtitle without language code in the
                                  file name, i.e. use .srt extension. Do not
                                  use this unless your media player requires
  -f, --force                     Force download even if a subtitle already
  -hi, --hearing-impaired         Prefer hearing impaired subtitles.
  -m, --min-score INTEGER RANGE   Minimum score for a subtitle to be
                                  downloaded (0 to 100).  [0<=x<=100]
  -w, --max-workers INTEGER RANGE
                                  Maximum number of threads to use.
  -z, --archives / -Z, --no-archives
                                  Scan archives for videos (supported
                                  extensions: .rar).  [default: z]
  -v, --verbose                   Increase verbosity.
  --help                          Show this message and exit.

subliminal cache

Usage: subliminal cache [OPTIONS]

  Cache management.

  --clear-subliminal  Clear subliminal's cache. Use this ONLY if your cache is
                      corrupted or if you experience issues.
  --help              Show this message and exit.

Provider Guide

This guide is going to explain how to add a Provider to subliminal. You are encouraged to take a look at the existing providers, it can be a nice base to start your own provider.


When starting a provider you should be able to answer to the following questions:

  • What languages does my provider support?
  • What are the language codes for the supported languages?
  • Does my provider deliver subtitles for episodes? for movies?
  • Does my provider require a video hash?

Each response of these questions will help you set the correct attributes for your Provider.

Video Validation

Not all providers deliver subtitles for Episode. Some may require a hash. The check() method does validation against a Video object and will return False if the given Video isn’t suitable. If you’re not happy with the default implementation, you can override it.


API keys must not be configurable by the user and must remain linked to subliminal. Hence they must be written in the provider module.

Per-user authentication is allowed and must be configured at instantiation as keyword arguments. Configuration will be done by the user through the provider_configs argument of the list_subtitles() and download_best_subtitles() functions. No network operation must be done during instantiation, only configuration. Any error in the configuration must raise a ConfigurationError.

Beyond this point, if an error occurs, a generic ProviderError exception must be raised. You can also use more explicit exception classes AuthenticationError and DownloadLimitExceeded.

Initialization / Termination

Actual authentication operations must take place in the initialize() method. If you need anything to be executed when the provider isn’t used anymore like logout, use terminate().

Caching policy

To save bandwidth and improve querying time, intermediate data should be cached when possible. Typical use case is when a query to retrieve show ids is required prior to the query to actually search for subtitles. In that case the function that gets the show id from the show name must be cached. Expiration time should be SHOW_EXPIRATION_TIME for shows and EPISODE_EXPIRATION_TIME for episodes.


To be able to handle various language codes, subliminal makes use of babelfish Language and converters. You must set the attribute languages with a set of supported Language.

If you cannot find a suitable converter for your provider, you can make one of your own.


The query() method parameters must include all aspects of provider’s querying with primary types.


A custom Subtitle subclass must be created to represent a subtitle from the provider. It must have relevant attributes that can be used to compute the matches of the subtitle against a Video object.

Score computation

To be able to compare subtitles coming from different providers between them, the get_matches() method must be implemented.


All possible uses of query(), list_subtitles() and download_subtitle() methods must have integration tests. Use vcrpy for recording and playback of network activity. Other functions must be unittested. If necessary, you can use unittest.mock to mock some functions.

API Documentation

If you are looking for information on a specific function, class or method, this part of the documentation is for you.



Supported archive extensions


Video extensions



Subtitle extensions











Refiners enrich a Video object by adding information to it.

A refiner is a simple function:

subliminal.refiners.refine(video, **kwargs)
  • video (Video) – the video to refine.
  • **kwargs – additional parameters for refiners.


subliminal.refiners.metadata.refine(video, embedded_subtitles=True, **kwargs)

Refine a video by searching its metadata.

Several Video attributes can be found:

  • resolution
  • video_codec
  • audio_codec
  • subtitle_languages

embedded_subtitles (bool) – search for embedded subtitles.


subliminal.refiners.tvdb.refine(video, **kwargs)

Refine a video by searching TheTVDB.


This refiner only work for instances of Episode.

Several attributes can be found:

  • series
  • year
  • series_imdb_id
  • series_tvdb_id
  • title
  • imdb_id
  • tvdb_id


subliminal.refiners.omdb.refine(video, apikey=None, **kwargs)

Refine a video by searching OMDb API.

Several Episode attributes can be found:

  • series
  • year
  • series_imdb_id

Similarly, for a Movie:

  • title
  • year
  • imdb_id



This module provides the default implementation of the compute_score parameter in download_best_subtitles() and download_best_subtitles().


To avoid unnecessary dependency on sympy and boost subliminal’s import time, the resulting scores are hardcoded here and manually updated when the set of equations change.

Available matches:

  • hash
  • title
  • year
  • country
  • series
  • season
  • episode
  • release_group
  • streaming_service
  • source
  • audio_codec
  • resolution
  • hearing_impaired
  • video_codec
  • series_imdb_id
  • imdb_id
  • tvdb_id




Expiration time for show caching


Expiration time for episode caching


Expiration time for scraper searches


The CacheRegion

Refer to dogpile.cache’s region configuration documentation to see how to configure the region


Subliminal uses click to provide a powerful CLI.





Antoine Bertin


Jan 20, 2023 2.1.0