Here's how it works:
* Audio Extraction: First, it extracts the audio from the video.
* Speaker Diarization: It then identifies the different speakers in the audio.
* Split Audio: The audio is split into smaller chunks based on the identified speakers.
* Speech to Text: Each chunk is transcribed into text.
* Combine ASR and Diarization: The transcriptions (from Automatic Speech Recognition) are combined with the diarization results to provide a structured, text-based dialogue for each identified speaker.
* Summarization: Finally, the dialogue is condensed into a summary for a quick overview.
The entire process is containerized to ensure seamless and efficient operation.
I'd love to get feedback or suggestions.
