Include option to add vocabulary for better transcription

[Jalaj-KT]

Problem statement

This pull request addresses the issue of suboptimal transcription quality when using Whisper for video dubbing. Users may experience bad transcription even after using large size models (for uncommon words like vishing, phising etc.)

Solution

The proposed solution introduces an option for users to input custom vocabulary. This enhancement aims to improve transcription accuracy, even with smaller models, by allowing specific words to be recognized more accurately.

Dependencies

No dependencies added.

[agung2001]

code change looks good, wish this can be merged @R3gm