| from transformers import AutoTokenizer, AutoModelForCausalLM | |
| from transformers import pipeline | |
| import gradio as gr | |
| from fastrtc import (ReplyOnPause, Stream, get_stt_model, get_tts_model) | |
| import numpy as np | |
| # messages = [ | |
| # {"role": "user", "content": "Who are you?"}, | |
| # ] | |
| # pipe = pipeline("text-generation", model="meta-llama/Llama-3.2-3B-Instruct") | |
| # pipe(messages) | |
| chatbot = pipeline("text-generation", model="microsoft/DialoGPT-medium") | |
| # tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-3.2-3B-Instruct") | |
| # model = AutoModelForCausalLM.from_pretrained("meta-llama/Llama-3.2-3B-Instruct") | |
| stt_model = get_stt_model() | |
| tts_model = get_tts_model() | |
| #stream = Stream(ReplyOnPause(echo), modality="audio", mode="send-receive") | |
| def echo(audio): | |
| prompt = stt_model.stt(audio) | |
| # response = sambanova_client.chat.completions.create( | |
| # model="Meta-Llama-3.2-3B-Instruct", | |
| # messages=[{"role": "user", "content": prompt}], | |
| # max_tokens=200, | |
| # ) | |
| # prompt = response.choices[0].message.content | |
| bot_response = chat_with_bot(prompt) | |
| audio_chunks = [] | |
| for audio_chunk in tts_model.stream_tts_sync(bot_response): | |
| audio_chunks.append(audio_chunk) | |
| return audio_chunks[0] | |
| # for audio_chunk in tts_model.stream_tts_sync(prompt): | |
| # yield audio_chunk | |
| def process_audio(audio_input): | |
| # audio_input is received as a Gradio Audio object, containing a tuple of (sample_rate, numpy array) | |
| sample_rate, audio_data = audio_input | |
| # Process audio through the stream | |
| processed_audio = echo((sample_rate, audio_data)) | |
| # Return processed audio to Gradio for output | |
| return sample_rate, processed_audio | |
| def chat_with_bot(user_input): | |
| # Generate a response from the chatbot model | |
| response = chatbot(user_input) | |
| return response[0]['generated_text'] | |
| # interface = gr.Interface( | |
| # fn=chat_with_bot, # Function to call for processing the input | |
| # inputs=gr.Textbox(label="Enter your message"), # User input (text) | |
| # outputs=gr.Textbox(label="Chatbot Response"), # Model output (text) | |
| # title="Chat with DialoGPT", # Optional: Add a title to your interface | |
| # description="Chat with an AI model powered by DialoGPT!" # Optional: Add a description | |
| # ) | |
| interface = gr.Interface( | |
| fn=process_audio, # The function to process audio | |
| inputs=gr.Audio(type="numpy"), # Microphone input (audio) | |
| outputs=gr.Audio(type="numpy"), # Audio output (processed) | |
| live=True # Make the processing live (if needed) | |
| ) | |
| interface.launch() |