AudioBookSync

Welcome to AudioBookSync!

If you have ever switched between reading a book and listening to the audiobook for that book, you might have found yourself trying to pinpoint the page that you are on in the book by jumping around the audiobook, listening to chunks of text and trying to triangulate where you are. These days are over.

AudioBookSync allows you to quickly and easily switch between a book and its associated audiobook to optimize your reading experience. When you are reading a book and want to switch over to the audiobook format, fire up AudioBookSync and type in a few words on the page so that our syncing algorithm can pinpoint where you are in the physical book and align this with the audiobook. AudioBookSync will give you the exact timestamp of where you should start listening in the audiobook!

Usage

To use AudioBookSync, simply clone the repository and run python AudioBookSync using the following arguments:

Note: Either --audio or --transcription_path must be supplied, but all other arguments are optional.

1. --audio:

   • Type: String
   • Description: Path to the audio file. Supported formats include mp3, mp4, mpeg, mpga, m4a, wav, and webm. If not specified, the model will use the transcription_path to search the transcription file for the search string.

2. --transcription_path:

   • Type: String (Default: None)
   • Description: Manually pass the path to the transcription file generated after transcription with Whisper. When supplied, the model will not run inference and will use the specified transcription file instead.

3. --language:

   • Type: String (Default: "en")
   • Description: Language code of the audio file. Default is set to 'en' for English.

4. --batch_size:

   • Type: Integer (Default: 16)
   • Description: Batch size for inference. Default is set to 16.

5. --custom_transcription_path:

   • Type: String (Default: None)
   • Description: Designate a custom path for the output file of the transcription. Default is the path to the audio file + SyncKey.json.

6. --model_type:

   • Type: String (Default: "tiny")
   • Description: Model type to use for inference. Possible values are 'tiny', 'base', 'small', 'medium', 'large', and 'large-v2'. Default is set to 'tiny'.

7. --transcription_only:

   • Type: Boolean flag (Default: False)
   • Description: Flag to indicate that only transcription is required. No ability to search the transcription will be provided when this flag is set to True.

8. --device:

   • Type: String (Default: "cuda")
   • Description: Device to use for inference. Possible values are 'cuda' and 'cpu'. Default is set to 'cuda'.