This project aims in transcribe either audio or video generating a SubRip file. If audio is selected, can also generate a video to be able to be read by most players.
In order to use this software, you must obtain a language and acustic Tensorflow model. You may either find both on the Internet or train them by yourself. Once downloaded, you need to download a Tensorflow model pretrained in the language of your need (or train one yourself) and place it inside of a folder named Languages in DeepSpeechUnity/Transcripter_Data/StreamingAssets.
- German.pbmm -> DeepSpeechUnity/Transcripter_Data/StreamingAssets/Languages/
- German.scorer -> DeepSpeechUnity/Transcripter_Data/StreamingAssets/Languages/
This project is tested using Aashish Agarwal's model which is wonderful and free. If you're using this software for transcripting from german I encourage you to download his model here https://github.com/AASHISHAG/deepspeech-german
In regard of the advance options, the ones selected as default come from Aashish Agarwal's flags.txt file which describes his selections when training, if using another model I recommend search for this kind of information in the place you found it and save changes.
This project is licensed under Apache Public License 2.0 and is based in Mozilla's DeepSpeech