suitable for concurrent voice chat bots using c# ?

[giannik]

Thank you for this project .

1. Would this be suitable for a voice chat bot application in asp.net core to handle the STT aspect with concurrent user calls for a voice chat bot ?
   2 . In terms of multi lingual support would it support othe languages apart from english . Im specifically interested in Greek langiage support.
   Thank you.

[TheAjaykrishnanR]

Thank you for this project .

1. Would this be suitable for a voice chat bot application in asp.net core to handle the STT aspect with concurrent user calls for a voice chat bot ?
   2 . In terms of multi lingual support would it support othe languages apart  from english . Im specifically interested in Greek langiage support.
   Thank you.

You are welcome !

1. The TTS part here is based on LlamaSharp which most likely supports concurrent generations with InferAsync(). LlamaCpp and llama python surely does so i dont see why it shouldnt, though i havent tried it myself. Now the SNAC decoder doesnt do async so would have to be run in its own thread, but it all depends on the implementation and whether the decoder runs on the server with LlamaSharp or on the client (this is much faster as you dont have to transmit audio to the client).

2. Orpheus does support many languages but i am not sure a Greek version exists, if not you will have to train it yourselves. I remember seeing in their issues people training and releasing other languages such as Turkish. There should be instructions there about dataset preparation and training.

[giannik]

@TheAjaykrishnanR thank you for the answers .
Need to spin up a machine with nvidia gpus to try it out . is there a windows machine configuration with an nvidia gpu setup you would recommed for testing ?
Thanks again.

[TheAjaykrishnanR]

@TheAjaykrishnanR thank you for the answers .
Need to spin up a machine with nvidia gpus to try it out . is there a windows machine configuration with an nvidia gpu setup you would recommed for testing ?
Thanks again.

I dont think you would need anything special, since if you are on windows all dependencies are managed by dotnet. With cuda 12, up and running it should be pretty straightforward.