Spaces:
Sleeping
Sleeping
Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,47 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import os
|
2 |
+
import gradio as gr
|
3 |
+
from groq import Groq
|
4 |
+
import whisper
|
5 |
+
from gtts import gTTS
|
6 |
+
import tempfile
|
7 |
+
|
8 |
+
# Set up Groq API key
|
9 |
+
os.environ['GROQ_API_KEY'] = 'GROQ_API-KEY'
|
10 |
+
groq_client = Groq(api_key=os.environ.get('GROQ_API_KEY'))
|
11 |
+
|
12 |
+
# Load Whisper model
|
13 |
+
whisper_model = whisper.load_model("base")
|
14 |
+
|
15 |
+
def process_audio(audio_file):
|
16 |
+
# Transcribe audio using Whisper
|
17 |
+
result = whisper_model.transcribe(audio_file)
|
18 |
+
user_text = result['text']
|
19 |
+
|
20 |
+
# Generate response using Llama 8b model with Groq API
|
21 |
+
chat_completion = groq_client.chat.completions.create(
|
22 |
+
messages=[
|
23 |
+
{
|
24 |
+
"role": "user",
|
25 |
+
"content": user_text,
|
26 |
+
}
|
27 |
+
],
|
28 |
+
model="llama3-8b-8192",
|
29 |
+
)
|
30 |
+
response_text = chat_completion.choices[0].message.content
|
31 |
+
|
32 |
+
# Convert response text to speech using gTTS
|
33 |
+
tts = gTTS(text=response_text, lang='en')
|
34 |
+
audio_file = tempfile.NamedTemporaryFile(delete=False, suffix='.mp3')
|
35 |
+
tts.save(audio_file.name)
|
36 |
+
|
37 |
+
return response_text, audio_file.name
|
38 |
+
|
39 |
+
# Create Gradio interface
|
40 |
+
iface = gr.Interface(
|
41 |
+
fn=process_audio,
|
42 |
+
inputs=gr.Audio(type="filepath"),
|
43 |
+
outputs=[gr.Textbox(label="Response"), gr.Audio(label="Response Audio")],
|
44 |
+
live=True
|
45 |
+
)
|
46 |
+
|
47 |
+
iface.launch()
|