Python: Extracting audio fragments from video

Goal: Develop a script for processing video files, extracting audio, optimizing its size, and splitting it into parts.

Deadlines: No deadlines, please suggest your timelines

Budget: Please suggest your budget for approval

1. Extracting audio from video

• Task: Extract the audio track from the video file.
• Video formats:
  • Supported input video files in the following formats: .mp4, .mkv, .avi, .mov.
• Audio format:
  • Save the audio file in AAC (.m4a) or MP3 (.mp3).
• Duration limitations for video: The input video can be up to 90 minutes long.
• Possible tools: 
  • FFmpeg. 
  • Ensure that the audio file is saved without loss of quality.

2. Limiting audio track size

• Task: The audio track must not exceed 25MB.
• Actions if size is exceeded:
  • If the file size exceeds 25MB, the audio file needs to be split into parts, each not exceeding 25MB.

3. Splitting the file into parts

• Task: Split the audio file into segments not exceeding 25MB, considering moments of silence.
• Splitting rules:
  • Splitting should occur at points of silence (speech inactivity) lasting at least 500 ms.
  • Add a "buffer" of 2s before and after each segment to prevent loss of speech.
• Example:
  • If a moment of silence occurs at the 10.00 second mark, the segment may end in the range of 8–12 seconds.
• Possible tools:
  • Use WebrtcVAD or Pyannote.audio to determine speech activity.
  • After identifying intervals of active speech, perform the audio file splitting using Pydub or FFmpeg.

4. Saving files

• Saving modes:
  • Locally:
    • If the local saving mode is selected, audio files (both full and fragments) are saved in a local folder.
  • In S3 storage:
    • If the S3 saving mode is selected, all files must be uploaded to the S3 bucket.
• Mode configuration:
  • Saving is determined by a parameter in the .env file

5. Preparing and deploying the code

• Task:
  • Prepare a ready script that:
    • Runs on a local machine and in Docker.
    • Processes the video file from start to finish, performing all the operations described above.
  • Deploy the code in the provided git repository, access will be granted after approval.
  • Format the code clearly, with comments for key operations.

If you need further clarification on tasks or resources, please let me know.