app.py

 # Install required libraries
import os
from io import BytesIO
import gradio as gr
from gtts import gTTS
from pydub import AudioSegment
import whisper
import openai

# Set up OpenAI API key
os.environ["OPENAI_API_KEY"] = "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"
openai.api_key = os.environ["OPENAI_API_KEY"]

# Initialize models
whisper_model = whisper.load_model("base")  # Load Whisper model

# Define the voice-to-voice workflow
def voice_to_voice(audio):
    # 1. Transcribe audio using Whisper
    transcription_result = whisper_model.transcribe(audio, fp16=False)
    user_input = transcription_result["text"]
    
    # 2. Get response from OpenAI's GPT
    response = openai.ChatCompletion.create(
        model="gpt-4",
        messages=[{"role": "user", "content": user_input}],
    )
    response_text = response.choices[0].message["content"]
    
    # 3. Convert LLM response to audio using gTTS
    tts = gTTS(text=response_text, lang="en")
    audio_fp = BytesIO()
    tts.write_to_fp(audio_fp)
    audio_fp.seek(0)
    
    # Convert gTTS output to a playable format using pydub
    audio_segment = AudioSegment.from_file(audio_fp, format="mp3")
    output_fp = BytesIO()
    audio_segment.export(output_fp, format="mp3")
    output_fp.seek(0)
    
    return response_text, output_fp

# Gradio interface
iface = gr.Interface(
    fn=voice_to_voice,
    inputs=gr.Audio(type="filepath"),
    outputs=[gr.Textbox(label="Transcription"), gr.Audio(label="Response Audio")],
    live=True,
    title="Real-Time Voice-to-Voice Chatbot",
    description="Speak into the microphone and get a spoken response from the chatbot.",
)

# Launch Gradio app
if __name__ == "__main__":
    iface.launch()