TTS that sounds less fake for Kimi

[cobaltspark]

Spent some time on this today and wanted to compare notes.

I've gone back and forth between vLLM and TensorRT-LLM for Phi. The interesting difference for me has been consumer GPUs.

What would you pick if you had to live with it for a few months?

For context, I'm not chasing leaderboard numbers. I mostly want the system to feel consistent day to day.

[visionforge]

Interesting, I hadn't tried that. I don't think your conclusion is far off. The biggest change for me was simplifying the stack instead of adding one more optimization layer.

[kernelpeak]

I don't think that's the whole story. I get why that seems right, but on GB200 the bottleneck felt elsewhere.

[bytegarden73]

Did a small comparison this afternoon. My result was mixed: first run looked bad, second run looked normal. The surprise was that memory pressure showed up before throughput collapsed.

[macropoint]

same here on my side.

[cobaltspark]

that fixed it for me too.

There are a few different conversations hiding inside this thread, so being explicit about the goal helps.

[mellowstream]

I hit almost the same issue last night.

[bytegarden73]

I hit almost the same issue last night.

What worked for me was pretty simple. On mini PC, the easiest win is often to back off the aggressive settings and see whether the problem survives. Once you know the plain path works, then add the faster path back in piece by piece.

[plainline]

I hit almost the same issue last night.

I'd approach it in a boring order. Before rebuilding anything, I'd test one clean baseline with Llama and compare from there. Not glamorous advice, but repeatable baselines save a lot of time.

[quiethouse]

A couple things I'd check first. If you're on TensorRT-LLM, I'd confirm the version first and then reduce variables until the behavior is obvious. Not glamorous advice, but repeatable baselines save a lot of time.