Spaces:
Running
Running
Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,64 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import os
|
2 |
+
import whisper
|
3 |
+
import gradio as gr
|
4 |
+
from groq import Groq
|
5 |
+
from TTS.api import TTS
|
6 |
+
|
7 |
+
# β
Set up Groq API Key
|
8 |
+
os.environ["GROQ_API_KEY"] = "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"
|
9 |
+
|
10 |
+
# β
Load OpenAI Whisper model
|
11 |
+
whisper_model = whisper.load_model("base")
|
12 |
+
|
13 |
+
# β
Function to Transcribe Speech to Text
|
14 |
+
def transcribe_audio(audio):
|
15 |
+
print("π Transcribing...")
|
16 |
+
result = whisper_model.transcribe(audio)
|
17 |
+
return result["text"]
|
18 |
+
|
19 |
+
# β
Function to Get Response from Groq API
|
20 |
+
def get_groq_response(prompt):
|
21 |
+
print("π€ Generating Response...")
|
22 |
+
client = Groq(api_key=os.environ["GROQ_API_KEY"])
|
23 |
+
chat_completion = client.chat.completions.create(
|
24 |
+
messages=[{"role": "user", "content": prompt}],
|
25 |
+
model="llama-3.3-70b-versatile"
|
26 |
+
)
|
27 |
+
return chat_completion.choices[0].message.content
|
28 |
+
|
29 |
+
# β
Function to Convert Text to Speech using Coqui TTS
|
30 |
+
def text_to_speech(text):
|
31 |
+
print("π Generating Speech...")
|
32 |
+
tts = TTS(model_name="tts_models/en/ljspeech/glow-tts")
|
33 |
+
output_file = "response.wav"
|
34 |
+
tts.tts_to_file(text=text, file_path=output_file)
|
35 |
+
return output_file
|
36 |
+
|
37 |
+
# β
Gradio Interface Function
|
38 |
+
def chatbot_pipeline(audio):
|
39 |
+
# Step 1: Convert Speech to Text
|
40 |
+
text = transcribe_audio(audio)
|
41 |
+
|
42 |
+
# Step 2: Get Response from Groq API
|
43 |
+
response = get_groq_response(text)
|
44 |
+
|
45 |
+
# Step 3: Convert Text Response to Speech
|
46 |
+
speech_file = text_to_speech(response)
|
47 |
+
|
48 |
+
return text, response, speech_file
|
49 |
+
|
50 |
+
# β
Build Gradio UI
|
51 |
+
interface = gr.Interface(
|
52 |
+
fn=chatbot_pipeline,
|
53 |
+
inputs=gr.Audio(type="filepath"), # Mic input
|
54 |
+
outputs=[
|
55 |
+
gr.Textbox(label="Transcribed Text"),
|
56 |
+
gr.Textbox(label="Chatbot Response"),
|
57 |
+
gr.Audio(label="Generated Speech")
|
58 |
+
],
|
59 |
+
title="π£οΈ Speech-to-Text AI Chatbot",
|
60 |
+
description="ποΈ Speak into the microphone β Get AI response β Listen to the reply!"
|
61 |
+
)
|
62 |
+
|
63 |
+
# β
Launch Gradio UI
|
64 |
+
interface.launch(share=True)
|