Use Case #3
Description
I am trying to find the use case for this, because I don't think I am using it correctly. I have a 3-tier setup: a front end, middleware (OpenAI TTS to Runpod serverless bridge), and a backend Runpod serverless instance handling inference. I initially thought I would LinaCodec encode on the back end and LinaCodec decode on the front end, but that would require some horsepower on the front end and some dependencies I could not control. So I was encoding and decoding on the backend, thinking the overall token count would be lower and that I could send the decoded stream upsampled to 48kHz, but I don't believe that's the correct use case either. On paper, this sounds really good, but I am not sure in my scenario if I can take advantage of it.