Update README.md

README.md

@@ -7,22 +7,24 @@ tags:
 - Text-to-Speech
 datasets:
 - CMU Arctic
+- Hi-Fi
 ---
 
 # TTS Models
 
 Here you can find models compatible with
-[balacoon_tts](https://pypi.org/project/balacoon_tts/) python package.
+[balacoon_tts](https://balacoon.com) python package.
 You can check interactive demo and models usage example in
 [balacoon/tts](https://huggingface.co/spaces/balacoon/tts) space.
 
 List of available models:
 
 - <mark>en_us_cmuarctic_pt_univenet.addon</mark> - en-US TTS trained on all 18
-  [CMU Arctic](http://festvox.org/cmu_arctic/) voices. This includes
-  examples of absolutely dashing accents, with "slp" voice being my
-  personal absolute favorite. Model consists of
-  [Parallel Tacotron](https://arxiv.org/abs/2010.11439)
-  and pre-trained [Univnet](https://github.com/mindslab-ai/univnet).
-  Model produces 24khz audio, although original dataset is unfortunately
-  16khz, so high frequencies are empty :(
+  [CMU Arctic](http://festvox.org/cmu_arctic/) voices.
+- <mark>en_us_hifi92_pt_univenet.addon</mark> - en-US TTS trained on
+  a single clean speaker(92) from [Hi-Fi dataset](https://arxiv.org/abs/2104.01497).
+
+Model names hints architecture used:
+
+- `pt_univenet` - combination of [Parallel Tacotron](https://arxiv.org/abs/2010.11439)
+  and pre-trained [Univnet](https://github.com/mindslab-ai/univnet).