msgflux/Kokoro-82M-streaming-onnx

Create config.json

Update README.md

Publish length-aware duration b96 checkpoint

Use English in the quickstart example

Publish validated adaptive b96 checkpoint

Update README.md

Update README.md

Delete onnx/kokoro-82m-streaming-b64-fp16.onnx

Delete onnx/kokoro-82m-streaming-b48-fp16.onnx

Delete onnx/kokoro-82m-streaming-b32-fp16.onnx

Delete onnx/kokoro-82m-streaming-b24-fp16.onnx

Delete onnx/kokoro-82m-streaming-b16-fp16.onnx

Document shifted tail-mask semantics

Document portable TensorRT-compatible graph

Publish final b96 and b128 portable checkpoints

Document stream boundary silence default

Document conservative b64 sentence splitting

Recommend b64 sentence streaming defaults

Recommend b48 as default streaming checkpoint

Document exact Kokoro ONNX export recipe

Update Kokoro streaming ONNX checkpoints and model card

Update Kokoro streaming ONNX checkpoints and model card

Update Kokoro streaming ONNX checkpoints and model card

Update Kokoro streaming ONNX checkpoints and model card

Update Kokoro streaming ONNX checkpoints and model card

Update Kokoro streaming ONNX checkpoints and model card

Update Kokoro streaming ONNX checkpoints and model card

Update Kokoro streaming ONNX checkpoints and model card

Tighten model card benchmark notes

Document b24 balanced default and b16 latency option

Recommend b16 streaming bucket

Update TensorRT server benchmark notes

Remove CUDA fusion note

Update install guidance and references

Document ONNX inputs and outputs

Recommend PCM response format

Simplify request example language option

Improve FastKokoro quickstart

Update model card and voice index

Update model card and voice index

Add fixed-bucket streaming ONNX exports

initial commit