Gooya experimental Persian TTS release.
GrainSpeech: Zitao Liang and Chang Gao, Apache-2.0; includes EfficientSpeech-derived code by Rowel Atienza (2023), Apache-2.0.
HiFi-GAN source and adapted generator: MIT, see licenses/HiFi-GAN-MIT.txt.
Negara v7.1 teacher/frontend: Apache-2.0, see licenses/Negara-Apache-2.0.txt.
Modifications: Persian vocabulary, 100-mel output, contextual encoder, duration supervision, speaker-adapted vocoder, frontend distillation, and standalone runtime.
Dataset attribution and limitations are in README.md; this package does not relicense source datasets.