Capability matrix
- Compute and format Pinned voice and weights · Intel i7-14700T.
- Local entry point WAV generation.
- Server Speech HTTP service. Text to speech
- Harness No recorded coding task. A dedicated Kokoro HTTP adapter exists. Check the selected voice, output audio and latency.
When to use it
Use the pinned Kokoro voice and model bundle and save a short WAV. On the reported i7-14700T run, 5.55 seconds of speech took a median 6.617 seconds to synthesize. Listen for clipped words and endings.
Configuration and tested scope · Publisher model card
Hardware and tuning
No comparable x86/ARM prefill and decode benchmark pair is recorded here. Full-checkpoint training, LoRA and other PEFT methods: no verified CKE recipe for this exact checkpoint in the reviewed evidence. For a trainable starter, use the FP32 and adapter recipes.