Lamma narrator voice (v1)

The on-device narrator voice of the Lamma party-games app. These files are a modified copy of Supertonic 3 by Supertone Inc., distributed under the BigScience OpenRAIL-M license included here as LICENSE.

Changes from the original (notice required by the license, §4c)

  • duration_predictor.onnx, text_encoder.onnx, vector_estimator.onnx: weights quantized to 8-bit integers (ONNX Runtime dynamic quantization, per channel, MatMul/Gemm/Conv).
  • vocoder.onnx: weights converted to 16-bit floats (inputs and outputs kept 32-bit).
  • tts.json, unicode_indexer.json and the voice styles F1.json, M1.json are unchanged.

Use restrictions

Use of these files, and of anything they generate, is subject to paragraph 5 and Attachment A of the license ("Use Restrictions"). Anyone you pass the model to must agree to the same restrictions. Speech it generates is computer-generated.

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for anventureslabs/lamma-narrator

Quantized
(36)
this model