Today
-
Add the PyTorch weights as model.safetensors, with the architecture config
Yesterday
-
Mock phone number in examples
-
Evaluation for step 190k
-
Model step 190k: Dicle fixed in all voices
-
Install the phonemizer from wite-tech/clarvia-phonemizer
-
Use wite-tech/clarvia-phonemizer instead of a bundled copy
-
Voices table without teacher references
-
Model (step 175k), config with 10 public voices, evaluation
-
Rename to clarvia-tts-tiny
-
Rename to clarvia-tts-tiny
-
Link the phonemizer repository
-
Licence line: Apache-2.0
-
Model card: using the phonemizer
-
Add the Turkish phonemizer and the inference script
-
Model card
-
initial commit