Spaces:
Sleeping
Sleeping
Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,59 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
from transformers import pipeline
|
3 |
+
import torch
|
4 |
+
import soundfile as sf
|
5 |
+
import numpy as np
|
6 |
+
import scipy.io.wavfile as wavfile
|
7 |
+
|
8 |
+
device = "cuda:0" if torch.cuda.is_available() else "cpu"
|
9 |
+
device = 'cpu'
|
10 |
+
print("Device:", device)
|
11 |
+
|
12 |
+
pipe_translate = pipeline("translation", model="Helsinki-NLP/opus-mt-fr-en", device=device)
|
13 |
+
pipe_tts = pipeline("text-to-speech", model="facebook/mms-tts-eng", device=device) # Better quality, way faster than bark
|
14 |
+
|
15 |
+
def get_translation(text):
|
16 |
+
return pipe_translate(text)[0]["translation_text"]
|
17 |
+
|
18 |
+
def get_audio(text):
|
19 |
+
speech = pipe_tts(text)
|
20 |
+
return speech["sampling_rate"], (speech["audio"]* 32767).astype(np.int16).T
|
21 |
+
|
22 |
+
with gr.Blocks() as demo:
|
23 |
+
input_text = gr.Textbox(
|
24 |
+
label="Input text",
|
25 |
+
info="Your text",
|
26 |
+
lines=3,
|
27 |
+
placeholder="Écrire le texte à traduire",
|
28 |
+
)
|
29 |
+
|
30 |
+
|
31 |
+
|
32 |
+
translation_button = gr.Button("Traduire...")
|
33 |
+
output_text = gr.Textbox(
|
34 |
+
label="Output text",
|
35 |
+
info="Your text",
|
36 |
+
lines=3,
|
37 |
+
placeholder="Votre traduction",
|
38 |
+
)
|
39 |
+
speech_button = gr.Button("Générer audio...")
|
40 |
+
translation_button.click(
|
41 |
+
get_translation,
|
42 |
+
inputs=[
|
43 |
+
input_text
|
44 |
+
],
|
45 |
+
outputs=[
|
46 |
+
output_text
|
47 |
+
],
|
48 |
+
)
|
49 |
+
speech_button.click(
|
50 |
+
get_audio,
|
51 |
+
inputs=[
|
52 |
+
output_text
|
53 |
+
],
|
54 |
+
outputs=[
|
55 |
+
gr.Audio(label="Output")
|
56 |
+
],
|
57 |
+
)
|
58 |
+
|
59 |
+
demo.launch()
|