v0.1

Author: ntt123

.gitignore

@@ -1,3 +1,4 @@
+test_data/
 assets/reinfo/
 # IDE files
 .idea

README.md

@@ -1,7 +1,9 @@
 A Vietnamese TTS
 ================
 
-Tacotron + WaveRNN for vietnamese dataset.
+Tacotron + WaveRNN for vietnamese datasets.
+
+A synthesized audio clip is at [assets/reinfo/clip.wav](assets/reinfo/clip.wav).
 
 Install
 -------
@@ -13,20 +15,38 @@ cd vietTTS
 pip3 install -e .
 ```
 
+
+Quick start using pretrained models
+----------------------------------
+```sh
+bash ./scripts/quick_start.sh
+```
+
+
 Download reinfo dataset
 -----------------------
 
 ```sh
 bash ./scripts/download_reinfo_dataset.sh
 ```
 
-Train Tacotron 
---------------
+
+Train duration model
+--------------------
 
 ```sh
-python3 -m vietTTS.tacotron.trainer
+python3 -m vietTTS.nat.duration_trainer
 ```
 
+
+Train acoustic model
+--------------------
+```sh
+python3 -m vietTTS.nat.acoustic_trainer
+```
+
+
+
 Train waveRNN
 -------------
 
@@ -39,5 +59,5 @@ Synthesize speech
 -----------------
 
 ```sh
-python3 -m vietTTS.synthesizer --text="####### hôm qua em tới trường #######" --output=clip.wav
+python3 -m vietTTS.synthesizer --use-nat --text="hôm qua em tới trường" --output=clip.wav
 ```