app.py(change_tts)

import gradio as gr
from huggingface_hub import InferenceClient
from transformers import pipeline
from datasets import load_dataset
import soundfile as sf
import torch
from asr import transcribe_auto  # ASR function

# Initialize Chat Model
chat_client = InferenceClient("Futuresony/future_ai_12_10_2024.gguf")

# Initialize Facebook TTS Model
tts_synthesizer = pipeline("text-to-speech", model="Futuresony/Output")

# Load Speaker Embeddings for TTS
embeddings_dataset = load_dataset("Matthijs/cmu-arctic-xvectors", split="validation")
speaker_embedding = torch.tensor(embeddings_dataset[7306]["xvector"]).unsqueeze(0)

def speech_to_chat(audio, history, system_message, max_tokens, temperature, top_p):
    # Step 1: Transcribe Speech to Text
    transcribed_text = transcribe_auto(audio)

    # Step 2: Generate Chat Response
    messages = [{"role": "system", "content": system_message}]
    
    for val in history:
        if val[0]:
            messages.append({"role": "user", "content": val[0]})
        if val[1]:
            messages.append({"role": "assistant", "content": val[1]})
    
    messages.append({"role": "user", "content": transcribed_text})
    
    response = ""
    for msg in chat_client.chat_completion(
        messages,
        max_tokens=max_tokens,
        stream=True,
        temperature=temperature,
        top_p=top_p,
    ):
        token = msg.choices[0].delta.content
        response += token

    # Step 3: Convert Chat Response to Speech
    speech = tts_synthesizer(response, forward_params={"speaker_embeddings": speaker_embedding})
    output_file = "generated_speech.wav"
    sf.write(output_file, speech["audio"], samplerate=speech["sampling_rate"])
    
    # Update Chat History
    history.append((transcribed_text, response))
    
    # Return transcribed text, chatbot response, generated speech, and updated history
    return transcribed_text, response, output_file, history

# Gradio Interface
with gr.Blocks() as demo:
    gr.Markdown("<h2 style='text-align: center;'>Real-time ASR → Chat → TTS</h2>")
    
    with gr.Row():
        audio_input = gr.Audio(source="microphone", type="filepath", label="🎤 Speak Here")
        transcribed_text_output = gr.Textbox(label="📝 Transcribed Text", interactive=False)
        chat_response_output = gr.Textbox(label="🤖 AI Response", interactive=False)
        audio_output = gr.Audio(label="🔊 AI Speech Output")

    submit_button = gr.Button("🎙️ Speak & Generate Response")

    system_msg = gr.Textbox(value="You are a friendly chatbot.", label="System Message")
    max_tokens = gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max New Tokens")
    temperature = gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature")
    top_p = gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p")
    chat_history = gr.State([])  # Store conversation history

    submit_button.click(
        fn=speech_to_chat,
        inputs=[audio_input, chat_history, system_msg, max_tokens, temperature, top_p],
        outputs=[transcribed_text_output, chat_response_output, audio_output, chat_history],
    )

# Run the App
if __name__ == "__main__":
    demo.launch()