Develop Real-Time OpenAI Whisper Speech-to-Text

Job ID: 36686870

Budget: $750 – $1,500 USD

We have a speech to text Windows application in C++ using OpenAI Whisper. However, there is no current functionality for Real-time streaming without using any "hackish/workaround" methods. This is because whisper is splitting the audio into 30 second chunks. We need this functionality invented so that we can use it in our speech to text program.

Requirements:
have a clear understanding of transformer models on a lower level meaning you can modify decoder/encoder blocks in whisper and you have a clear understanding of how they work
you should be able to make a low level implementation if needed taking advantage of SIMD instruction sets in cpp/c