This message was deleted.
# sdk-rust
s
This message was deleted.
b
I’ve done the exact same thing using our Python SDK here
🙌 1
Should be very similar using the Rust SDK
b
Just seeing this, I came up with a super hacky way using the whisper-rs crate (https://github.com/cs50victor/kitt_2/blob/main/lkgpt/src/stt.rs), but the performance is really poor. I'll try using your example
b
Hey, I may be because you’re doing the inference inside the tokio runtime
You should not block inside the runtime
You can use spawn_blocking and await it
b
hmm. interesting. thanks I'll try that. I was wondering if there is some form of limit to the number of threads the server can spawn without being too computationally intensive?
just tested the whisper example and I got this segmentation fault