Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -8,67 +8,7 @@ from dotenv import load_dotenv
|
|
8 |
load_dotenv()
|
9 |
image_API="https://api-inference.huggingface.co/models/stabilityai/stable-diffusion-2"
|
10 |
translation_API="https://api-inference.huggingface.co/models/facebook/nllb-200-distilled-1.3B"
|
11 |
-
|
12 |
-
|
13 |
-
HF_TOKEN=os.getenv("HF_Token")
|
14 |
-
login(HF_TOKEN)
|
15 |
-
headers={"Authorization":f"Bearer {HF_TOKEN}"}
|
16 |
-
|
17 |
-
|
18 |
-
def query_whisper(audio_path):
|
19 |
-
with open(audio_path,"rb") as audio:
|
20 |
-
data=audio.read()
|
21 |
-
response=requests.post(whisper_API,headers=headers,data=data)
|
22 |
-
return response.json()
|
23 |
-
|
24 |
-
def query_translation(text):
|
25 |
-
max_retries=5
|
26 |
-
delay=10
|
27 |
-
payload={"inputs":text,"parameters":{"src_lang":"ta_Taml","tgt_lang":}}
|
28 |
-
for attempt in range(max_retries):
|
29 |
-
response=requests.post(translation_API,headers=headers,json=payload)
|
30 |
-
result= response.json()
|
31 |
-
if isinstance(result,list) and len(result)>0:
|
32 |
-
return {"translated_text":result[0]["translation_text"]}
|
33 |
-
elif isinstance(result,dict) and "translation_text" in result:
|
34 |
-
return {"translated_text":result["translation_text"]}
|
35 |
-
time.sleep(delay)
|
36 |
-
return {"error":"Translation API failed","response":result}
|
37 |
-
|
38 |
-
def query_image(prompt):
|
39 |
-
max_retries=5
|
40 |
-
delay=10
|
41 |
-
payload={"inputs":prompt}
|
42 |
-
for attempt in range(max_retries):
|
43 |
-
response=requests.post(image_API,headers=headers,json=payload)
|
44 |
-
if response.status_code==200 and response.get.headers('Content-Type','').startswith('image/'):
|
45 |
-
image_path='generated_image.png'
|
46 |
-
with open(image_path,'wb') as f:
|
47 |
-
f.write(response.content)
|
48 |
-
return image_path
|
49 |
-
time.sleep(delay)
|
50 |
-
return None
|
51 |
-
|
52 |
-
def process_audio(audio_path):
|
53 |
-
if not audio_path:
|
54 |
-
return None
|
55 |
-
try:
|
56 |
-
transcription=query_whisper(audio_path)
|
57 |
-
tamil_text=transcription.get("text","Transcription error")
|
58 |
-
translation=query_translation(tamil_text)
|
59 |
-
translated_text=translation.get("translated_text","Translation error")
|
60 |
-
image_path=query_image(translated_text)
|
61 |
-
return tamil_text,translated_text,image_path
|
62 |
-
except Exception as e:
|
63 |
-
return None,str(e),None
|
64 |
-
|
65 |
-
iface=gradio.Interface(
|
66 |
-
fn=process_audio,
|
67 |
-
inputs=gradio.Audio(type="filepath",label="Upload Audio"),
|
68 |
-
outputs=[
|
69 |
-
gradio.Textbox(label="Tamil Text"),
|
70 |
-
gradio.Textbox(label="English Translation"),
|
71 |
-
gradio.Image(label="Generated Image")
|
72 |
],
|
73 |
title= title="Speech-to-Image Generation"
|
74 |
)
|
|
|
8 |
load_dotenv()
|
9 |
image_API="https://api-inference.huggingface.co/models/stabilityai/stable-diffusion-2"
|
10 |
translation_API="https://api-inference.huggingface.co/models/facebook/nllb-200-distilled-1.3B"
|
11 |
+
… gradio.Image(label="Generated Image")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
12 |
],
|
13 |
title= title="Speech-to-Image Generation"
|
14 |
)
|