2 days ago
-
Add demo video
3 days ago
-
Model card: remove architecture and training details
4 days ago
-
Model card: point the license note to Acknowledgements
-
Model card: License section moved to the top
-
Model card: licence section (Apache-2.0, free to use; built on open-source components)
5 days ago
-
CPU inference: --threads option and a CPU section in the README
6 days ago
-
Quick start: inference script, mithi_tts package, requirements; README updated
-
Banner: layout 2 with language names kept clear of the logo
-
Banner: layout 2
-
Banner: larger language names across the whole banner
-
Banner: v1 Flash beside the wordmark, languages scattered in different sizes, no voices line
-
Add banner (wordmark + sound rangoli) and wordmark to the model card
-
Model card: trim internal details, shorter limitations, acknowledge Maya2-native
-
English sample: 'see them perform live' (live read as the adjective /laɪv/)
-
Model card: samples as two tables (conversational; news and education)
-
Add 44 samples (11 languages x 2 sentences x 2 voices) with a table in the model card
-
Model card: voices section (one male, one female speaker from Maya2-native inference)
-
Mithi v1 Flash: bf16 weights, model card, licence
-
initial commit
