Help with Audio / Speech

[sparrowhall]

Thought I'd post a small recipe because a lot of threads ask similar things.

Just sharing a setup that ended up working reasonably well: Phi, SGLang, Threadripper box, with a lot of attention on prefix caching.

If someone wants the exact settings I can clean them up and post them.

[emberrunner]

I'd approach it in a boring order. If you're on PyTorch, I'd confirm the version first and then reduce variables until the behavior is obvious. Once you know the plain path works, then add the faster path back in piece by piece.

[sparrowcraft]

I tested this tonight. My result was mixed: first run looked bad, second run looked normal.

[cudapoint]

A couple things I'd check first. Before rebuilding anything, I'd test one clean baseline with Kimi and compare from there. A lot of these threads end up being one bad config plus a pile of conclusions drawn from it.

[dockridge]

Good point. I ended up in about the same place using MLX.

[vectorfield]

I'd start with the simpler stack.

[vramridge]

Thanks, that helps. The pattern around router policy has been pretty similar for me.

[amberhall]

My first guess would be this: If you're on llama.cpp, I'd confirm the version first and then reduce variables until the behavior is obvious. A lot of these threads end up being one bad config plus a pile of conclusions drawn from it.

[sanescope]

Need a little more detail here. Do you see the same thing with a smaller batch or shorter prompt?

[vramforge]

Need a little more detail here. Do you see the same thing with a smaller batch or shorter prompt?

My first guess would be this: On 4090, the easiest win is often to back off the aggressive settings and see whether the problem survives. A lot of these threads end up being one bad config plus a pile of conclusions drawn from it.