Update README.md

README.md

@@ -8,6 +8,7 @@ tags:
 - matcha-tts
 ---
 # Matcha-TTS CommonVoice EN001 
+[Github Demo](https://akjava.github.io/Matcha-TTS-Japanese/matcha_tts_speak_en001.html)
 
 ## Source Audio
 https://commonvoice.mozilla.org/en/datasets
@@ -16,13 +17,13 @@ Common Voice Corpus 1
 I called audios 42da7f26(head-audio-id)_290(files) EN001
 (No plan to include audios in this repo)
 ## Any Good point?
-LJSpeech is much better quality,but it's female voice.
+LJSpeech is much better quality,but it's female voice.This one is men.
 
 VCTK 107 voices are similar quality,but that is ODC-By License.
 
 This audio is just under MIT more easy to continue training or something.
 
-however I recommend you use VCTK,ODC-By License is not so problem.I'm going to create new voices with this.
+however I recommend you use VCTK,ODC-By License is not so problem.I'm going to create new voices with this in future.
 ## How to Train
 Train with IPA text(this folk)
 https://github.com/akjava/Matcha-TTS-Japanese
@@ -39,6 +40,9 @@ Sadly I lost between 3599 - 4499 checkpoints.I'm sorry.
 see Training metrics
 ### ONNX
 
+[github codes](https://github.com/akjava/Matcha-TTS-Japanese/tree/main/examples) - see sourcecode
+[github Page](https://akjava.github.io/Matcha-TTS-Japanese/) - Test Onnx Example
+
 onnx simplified loading speed is now 1.5 times faster.
 ```
 from onnxsim import simplify