Alma, a Swedish (sv_SE) Piper voice, medium quality, packaged to run with speaches (model.onnx + config.json, the same layout as the speaches-ai/piper-* repos).

Unchanged copy of sv_SE-alma-medium.onnx and its config from yeagersthlm/piper-voice-sv-alma by Daniel Nylander (retrained for 10,000 epochs; inference defaults noise_scale 0.8, noise_w 0.6, length_scale 0.9). Trained on the NST Swedish speech corpus (Språkbanken). License: CC BY 4.0, as the original.

Downloads last month
22
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support