LLMwithvoice.py

import requests
from pydub import AudioSegment
from io import BytesIO
import gradio as gr

# Hugging Face API URLs
API_URL_ROBERTA = "https://api-inference.huggingface.co/models/deepset/roberta-base-squad2"
API_URL_TTS = "https://api-inference.huggingface.co/models/espnet/english_male_ryanspeech_tacotron"

# Function to query the RoBERTa model
def query_roberta(api_token, prompt, context):
    payload = {
        "inputs": {
            "question": prompt,
            "context": context
        }
    }
    headers = {"Authorization": f"Bearer {api_token}"}
    response = requests.post(API_URL_ROBERTA, headers=headers, json=payload)
    try:
        response.raise_for_status()  # Raise an error for bad responses
        return response.json()
    except requests.exceptions.HTTPError as e:
        return {"error": f"HTTP error occurred: {e}"}
    except ValueError as e:
        return {"error": f"Value error occurred: {e}"}
    except Exception as e:
        return {"error": f"An unexpected error occurred: {e}"}

# Function to generate speech from text using ESPnet TTS
def generate_speech(api_token, answer):
    payload = {
        "inputs": answer,
    }
    headers = {"Authorization": f"Bearer {api_token}"}
    response = requests.post(API_URL_TTS, headers=headers, json=payload)
    try:
        response.raise_for_status()  # Raise an error for bad responses
        audio = response.content
        audio_segment = AudioSegment.from_file(BytesIO(audio), format="wav")
        audio_file = BytesIO()
        audio_segment.export(audio_file, format="wav")
        audio_file.seek(0)
        return audio_file
    except requests.exceptions.HTTPError as e:
        print(f"HTTP error occurred: {e}")
        return None
    except Exception as e:
        print(f"An unexpected error occurred: {e}")
        return None

# Function to interface with Gradio
def gradio_interface(api_token, context, prompt):
    answer = query_roberta(api_token, prompt, context)
    if 'error' in answer:
        return answer['error'], None
    answer_text = answer.get('answer', 'No answer found')
    audio_file = generate_speech(api_token, answer_text)
    return answer_text, audio_file