The current project structure looks like this:
.
├── docs
│ ├── code_documentation.adoc
│ ├── compiling.adoc
│ ├── installing_dependencies.adoc
│ ├── usage.adoc
│
├── install
│ ├── build.sh
│
├── libs
│ ├── lib_co-oCCur
│ ├── lib_ext
│
├── src
| |---utils
| | |---align_fingerprints.cpp
| | |---align_fingerprints.h
| | |---base64_string.cpp
| | |---base64_string.h
| | |---dactylogram.cpp
| | |---dactylogram.h
| |
| |---main.cpp
| |---subtitle_resync.cpp
| |---subtitle_resync.h
|
|---.gitignore
|---CMakeLists.txt
|---LICENSE.md
|---README.mdThis documentation covers the description of all the source files, the used methods,
variables which will help anyone understand what’s going on under the hood.
All the files can be found out inside the src directory.
All the classes and the code is under the namespace Subtitle_Resync. Most of the methods are documented here,
leaving a few helper methods which are properly named in camelCase and are easy to understand.
Anyway, if you have any method that I have missed, or better you want to contribute feel free to open a PR.
These files trigger everything. The names/paths of the files to be used, the audio file(s) and the subtitle file, is
passed to the Tool that is being used via the method init()
-
Getters and Setters!
std::string getToolName() const; void setTool(std::string ToolName); void setOriginalAudio(std::string OriginalAudioName); void setModifiedAudio(std::string ModifiedAudioName); void setOriginalSubtitle(std::string OriginalSRTName);
-
The triggering method present inside the
Subtitle_Resyncclass./* * Based on the Tool to be used, it creates an object * of the selected tool and calls the the required * methods for the further working. * * @params: None * Return value: None */ void Subtitle_Resync::Subtitle_Resync::init();
These files are responsible for the handling of the audio files, storing the data, generation of the audio fingerprints
using chromaprint.
/* * The audio file(.wav) is read in and the data inside it is * stored as a vector *int16_t* data. * * @params: None * Return value: None */ void Subtitle_Resync::Dactylogram::readAudio();
/* * The internal parameters required for chromaprint to work * are set, the required functions are called resulting the * generation of audio fingerprints of a complete audio file. * * @params: None * Return value: None */ void Subtitle_Resync::Dactylogram::igniteChromaprint();
/* * A public method to retrieve all the fingerprints generated and * store them in the form in which the are stored later. * * @params: None * Return value: None */ void Subtitle_Resync::Dactylogram::collectFingerprints();
/* * A getter used to get all the audio fingerprints for the * concerned audio file. * * @params: None * Return value: All the fingerprints, a vector of type uint32_t */ std::vector<uint32_t> Subtitle_Resync::Dactylogram::getAllFingerprints();
/* * Returns the length of the audio in seconds. * * @params: None * Return value: length in seconds, <type> double */ double Subtitle_Resync::Dactylogram::getAudioLength_secs();
/* * Return the length of the audio in milliseconds. * * @params: None * Return value: length in milliseconds, <type> long int */ long int Subtitle_Resync::Dactylogram::getAudioLength_ms();
Using the generated audio fingerprints, we use the class AlignFP and it’s methods to
make the subtitles Subtitle_Resync with the audiovisual content.
Here’s what this class' methods and fields are:
// Internally, the fingerprint is an array of 32-bit integer. // These integers are hashes of some arbitrary audio // features over a short period of time. There is a new hash // every *chroma_factor* seconds and each hash covers the 2.6 // seconds of audio constexpr static double chroma_factor = (4096.0 / 3) / 11025.0
/* * Finds the regions in the asked duration of the two audio files. * * @params: seconds - From where to find the matchin regions, <type> double * @params: offset - The offset determined using *align_fingerprints()*, <type> int * Return value: Returns the score for the first segment, <type> int */ int Subtitle_Resync::AlignFP::matching_regions(double seconds, int offset, bool local);
/* * Compares the fingerprints of the audio files, specifically * the a sgement of the fingerprints of the reference/original audio is compared against * the complete modified audio. Finds whether there is match, a local match or no match at all. * * @params: fp1_seg - reference to a vector containing the segment of audio fingerprints * to be used for the alignment. * Return value: Returns an integer among 1, 2 & 3. * 1 = MATCH * 2 = LOCAL_MATCH * 3 = NO_MATCH */ int Subtitle_Resync::AlignFP::align_fingerprints(const std::vector<uint32_t > &fp1_seg, bool print);
/* * Starting from t = 0, the beginning of the audio files it finds the * duration of the first segment and it decides whether it is a commercial or a part of the content. * * @params: None * Return value: None */ void Subtitle_Resync::AlignFP::ground_zero();
/* * Finds the duration next content segment in the original file. * * @params: i - index of the current segment * Return value: None */ void Subtitle_Resync::AlignFP::next(int i);
/* * Finds the duration of the next commercial in the original file. * * @params: i - index of the current segment. * Return value: None */ void Subtitle_Resync::AlignFP::next_comm(int i);
/* * A method knits everything together and calls the appropriate methods * to finds all the segments in the audio. * * @params: None * Return value: None */ void Subtitle_Resync::AlignFP::segment_it();
/* * A method that calls all the required methods, once the segments of the audio are * found, uses the methods of *SiftSRT* to adjust the timings of the subtitles resulting * in an in-sync subtitle file. * * @params: None * Return value: None */ void Subtitle_Resync::AlignFP::brum_brum()