SagyHarpazGong

@nithinraok Thanks for the response, unfortunately I can't share the audio but I can share the clustering parameters: ``` clustering: parameters: oracle_num_speakers: False # If True, use num of speakers...

@sivanantha321 Thanks for your response. I tried basically all ray versions until 2.7.1 and only 2.7.1 or lower are work. from ray version 2.8 the method "deploy" of the Deployment...

Is there any update for this issue? [#3668](https://github.com/kserve/kserve/pull/3668) is still not merged

Hi @nithinraok , thanks for your response. What ai meant is that there is a for loop that run the segmentation and the embedding extraction. But as I saw the...

Thanks @tango4j yes I also think so

Of course but the msdd use them so if I disable the saving of the pkl files I get FileNotFoundError

I want to use the msdd diarization

@tango4j I checked as well and still slow, I'm really suspect that the reason for the slow inference is the utilization of the CPUs, most of the inference time the...

What CUDA settings should I need to check? By increasing the batch_size the memory used in the GPU is almost on the maximum and still it's slow

@tango4j not at all