Roh commited on
Commit
996e356
1 Parent(s): 288ad42

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +16 -1
README.md CHANGED
@@ -13,7 +13,22 @@ widget:
13
  ## RyanSpeech model (based on ESPnet2)
14
 
15
  ### `espnet/english_male_ryanspeech_fastspeech2`
16
- This model was trained by [Rohola Zandie](https://scholar.google.com/citations?user=xv0jIe0AAAAJ&hl=en) using ryanspeech recipe in [espnet](https://github.com/espnet/espnet/).
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
17
 
18
 
19
  ## Download the dataset
13
  ## RyanSpeech model (based on ESPnet2)
14
 
15
  ### `espnet/english_male_ryanspeech_fastspeech2`
16
+ This model was trained by [Rohola Zandie](https://scholar.google.com/citations?user=xv0jIe0AAAAJ&hl=en) using ryanspeech recipe in [espnet](https://github.com/espnet/espnet/). For the best results you need to download the vocoder separately from [here](https://drive.google.com/file/d/10GYvB_mIKzXzSjD67tSnBhknZRoBjsNb/view?usp=sharing) and then use the following code:
17
+
18
+ ```
19
+
20
+ from espnet2.bin.tts_inference import Text2Speech
21
+ from scipy.io.wavfile import write
22
+
23
+ model = Text2Speech.from_pretrained(
24
+ model_file="espnet/english_male_ryanspeech_fastspeech2",
25
+ vocoder_file="path_to_vocoder/train_nodev_parallel_wavegan.v1.long/checkpoint-1000000steps.pkl"
26
+ )
27
+
28
+ output = model("This is a simple test.")
29
+
30
+ write("x.wav", 22050, output['wav'].numpy())
31
+ ```
32
 
33
 
34
  ## Download the dataset