Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -39,18 +39,17 @@ def load_model(model_path, hps):
|
|
39 |
#hps = utils.get_hparams_from_file("configs/vctk_base.json")
|
40 |
hps = utils.get_hparams_from_file("wa_graphemes/config.json")
|
41 |
|
42 |
-
# Define a dictionary to store the model paths for each tab
|
43 |
model_paths = {
|
44 |
"Graphemes": "wa_graphemes/G_258000.pth"
|
45 |
}
|
46 |
|
47 |
-
# Load the
|
48 |
net_g = load_model(model_paths["Graphemes"], hps)
|
49 |
|
50 |
def tts(text, speaker_id, tab_name):
|
51 |
global net_g
|
52 |
net_g = load_model(model_paths[tab_name], hps)
|
53 |
-
sid = torch.LongTensor([speaker_id]) # speaker
|
54 |
stn_tst = get_text(text, hps)
|
55 |
|
56 |
with torch.no_grad():
|
@@ -77,6 +76,13 @@ with app:
|
|
77 |
# First Text to Speech (TTS) for Walloon
|
78 |
Based on VITS (https://github.com/jaywalnut310/vits).
|
79 |
Write the text in graphemes. For faster inference speed it is recommended to use short sentences.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
80 |
"""
|
81 |
)
|
82 |
with gr.Tabs():
|
@@ -84,11 +90,13 @@ with app:
|
|
84 |
|
85 |
gr.Markdown(
|
86 |
"""
|
87 |
-
|
88 |
-
| Input Text | Speaker |
|
89 |
-
|
90 |
-
|
|
91 |
-
| Li bijhe et l’ solea estént ki s’ margayént po sawè kî çki, des deus, esteut l’ pus foirt. Mins ç’ côp la, la k’ i veyèt on tchminåd k' arivéve pyim piam, dins on bea noû tchôd paltot. | Male |
|
|
|
|
|
92 |
"""
|
93 |
)
|
94 |
|
|
|
39 |
#hps = utils.get_hparams_from_file("configs/vctk_base.json")
|
40 |
hps = utils.get_hparams_from_file("wa_graphemes/config.json")
|
41 |
|
|
|
42 |
model_paths = {
|
43 |
"Graphemes": "wa_graphemes/G_258000.pth"
|
44 |
}
|
45 |
|
46 |
+
# Load the model
|
47 |
net_g = load_model(model_paths["Graphemes"], hps)
|
48 |
|
49 |
def tts(text, speaker_id, tab_name):
|
50 |
global net_g
|
51 |
net_g = load_model(model_paths[tab_name], hps)
|
52 |
+
sid = torch.LongTensor([speaker_id]) # speaker ID
|
53 |
stn_tst = get_text(text, hps)
|
54 |
|
55 |
with torch.no_grad():
|
|
|
76 |
# First Text to Speech (TTS) for Walloon
|
77 |
Based on VITS (https://github.com/jaywalnut310/vits).
|
78 |
Write the text in graphemes. For faster inference speed it is recommended to use short sentences.
|
79 |
+
|
80 |
+
The quality of the results varies between male and female voice due to the limited data.
|
81 |
+
For better results with female voice use the phonemes based model.
|
82 |
+
|
83 |
+
https://huggingface.co/spaces/Pipe1213/VITS_Walloon_Phonemes
|
84 |
+
|
85 |
+
Hint: Some sample texts are available at the bottom of the web site.
|
86 |
"""
|
87 |
)
|
88 |
with gr.Tabs():
|
|
|
90 |
|
91 |
gr.Markdown(
|
92 |
"""
|
93 |
+
## Examples
|
94 |
+
| Input Text | Speaker |
|
95 |
+
|------------|---------|
|
96 |
+
| Portant, c' est l' seu ki n' doereut nén fé rire di lu, a mi idêye. | Female |
|
97 |
+
| Li bijhe et l’ solea estént ki s’ margayént po sawè kî çki, des deus, esteut l’ pus foirt. Mins ç’ côp la, la k’ i veyèt on tchminåd k' arivéve pyim piam, dins on bea noû tchôd paltot. | Male |
|
98 |
+
| Ci fourit co l' bedot les cåzes ca, a on moumint, li Ptit Prince mi dmanda yåk, come onk k' est so dotance, tot d' on côp | Female |
|
99 |
+
| Li Ptit Prince, da Antoenne di Sint-Spuri, ratourné e walon pa Lorint Enchel | Female |
|
100 |
"""
|
101 |
)
|
102 |
|