A browser-ready ONNX export of the German Kokoro-82M fine-tune by Software Mansion, used by the Mark to Read Chrome extension.
| File | Precision | Size | Intended runtime |
|---|---|---|---|
onnx/model.onnx |
fp32 | 326 MB | ONNX Runtime Web, WebGPU |
onnx/model_quantized.onnx |
dynamic uint8, per channel | 93 MB | ONNX Runtime Web, WASM / CPU |
voices/df_anna.bin |
float32 [510, 256] |
0.5 MB | voice style ("Anna", female) |
Same as the official Kokoro ONNX models:
input_ids int64 [1, n] (Kokoro v1.0 vocabulary, token 0 at both ends), style float32 [1, 256], speed float32 [1]waveform float32 (24 kHz mono), duration int64voice[phoneme_count - 1]Input phonemes are IPA in Kokoro's notation; the matching German G2P is Phonemis (MIT).
finetunes/kokoro_german_converted.pth from
software-mansion/react-native-executorch-kokoro
(revision 3744b57964eab7df6e8c48f0b84badb29e14df07, Apache-2.0).parametrizations.weight.original0/1
format; these were mapped back to weight_g/weight_v so that every weight loads (the stock
loader silently skips them).kokoro 0.9.4 (disable_complex=True), PyTorch 2.5.1, opset 17, with three
graph simplifications that are exact for batch size 1:
packed sequences removed, InstanceNorm written as mean/variance, and the all-false masks and
identity zero-padding copies removed. Without these, ONNX Runtime's WebGPU backend produces
wrong durations and garbled audio.model_quantized.onnx: onnxruntime.quantization.quantize_dynamic (QUInt8, per channel;
MatMul, Gemm, Conv, ConvTranspose, LSTM).Against the independent Kokoro KMP German export (CPU) and against each other, using log-spectrogram correlation:
Apache-2.0. Kokoro-82M by hexgrad; German fine-tune by Software Mansion; conversion by S3P Studios.
onnx/model.onnx 326 MBonnx/model_quantized.onnx 93 MBvoices/df_anna.bin 522.2 kBhexgrad/Kokoro-82M