Video Speech Check

Builds

Functionalities:

• speech to text recognition and transcribe
• check existence of desired and prohibited phrases

UI

Command line application

1. build/publish Visprech.Cmd project or download latest release
2. run the built executable
   • if not exists ffmpeg will be download (needed for audio extraction)
   • if not exists Whisper ggml will be downloaded (needed for transcript creation)
3. the app.conf file should be created with default configuration
4. edit phrases in app.conf. Add phrases that are desired and/or prohibited. DesiredPhrases and ProhibitedPhrases respectively.
5. run the executable against video file (audio file will work too).
   • Run it in the command line. E.g. VideoSpeechCheck.exe <file to check>
   • Drag-and-Drop a video file on VideoSpeechCheck.exe

Used resources:

• ffmpeg to extract audio stream for speech to text process
• Whisper.net (OpenAI Whisper) to transcribe speech to text
• Diacritics.NET used to remove diacritics before phrases check
• Fastenshtein used to compare phrases by Levenshtein distance

Planned improvements

• some more user friendly UI