Supertonic 3 for ExecuTorch

Supertone/supertonic-3 exported to ExecuTorch 1.5.1 for XNNPACK, the CPU backend. For details on the model itself, see the original model card.

file weights size
xnnpack_int8/supertonic.pte int8 (GPTQ), recommended 142 MB
xnnpack_fp32/supertonic.pte fp32 398 MB

Each .pte holds all four stages of the model: duration_predictor, text_encoder, vector_estimator and vocoder. Beside each one, ops.txt lists the kernels it needs from ExecuTorch, and metadata.json records how it was exported. export/ holds the scripts that produced the files.

License

OpenRAIL-M; see LICENSE. The use restrictions in Attachment A apply to these files and to anything derived from them. The .pte files are modified versions of Supertone's weights. Copyright (c) 2026 Supertone Inc. See NOTICE.

Downloads last month
2
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for lee-futo/supertonic-3-executorch

Quantized
(33)
this model