app.py

import os

import numpy as np

import gradio as gr
import requests
from genai_chat_ai import AI,create_chat_session

import torch
from typing import Any, Callable, Optional, Tuple, Union,Iterator
import numpy as np
import torch.nn as nn # Import the missing module



def _inference_forward_stream(
        self,
        input_ids: Optional[torch.Tensor] = None,
        attention_mask: Optional[torch.Tensor] = None,
        speaker_embeddings: Optional[torch.Tensor] = None,
        output_attentions: Optional[bool] = None,
        output_hidden_states: Optional[bool] = None,
        return_dict: Optional[bool] = None,
        padding_mask: Optional[torch.Tensor] = None,
        chunk_size: int = 32,  # Chunk size for streaming output
    ) -> Iterator[torch.Tensor]:
        """Generates speech waveforms in a streaming fashion."""
        if attention_mask is not None:
            padding_mask = attention_mask.unsqueeze(-1).float()
        else:
            padding_mask = torch.ones_like(input_ids).unsqueeze(-1).float()



        text_encoder_output = self.text_encoder(
            input_ids=input_ids,
            padding_mask=padding_mask,
            attention_mask=attention_mask,
            output_attentions=output_attentions,
            output_hidden_states=output_hidden_states,
            return_dict=return_dict,
        )
        hidden_states = text_encoder_output[0] if not return_dict else text_encoder_output.last_hidden_state
        hidden_states = hidden_states.transpose(1, 2)
        input_padding_mask = padding_mask.transpose(1, 2)

        prior_means = text_encoder_output[1] if not return_dict else text_encoder_output.prior_means
        prior_log_variances = text_encoder_output[2] if not return_dict else text_encoder_output.prior_log_variances

        if self.config.use_stochastic_duration_prediction:
            log_duration = self.duration_predictor(
                hidden_states,
                input_padding_mask,
                speaker_embeddings,
                reverse=True,
                noise_scale=self.noise_scale_duration,
            )
        else:
            log_duration = self.duration_predictor(hidden_states, input_padding_mask, speaker_embeddings)

        length_scale = 1.0 / self.speaking_rate
        duration = torch.ceil(torch.exp(log_duration) * input_padding_mask * length_scale)
        predicted_lengths = torch.clamp_min(torch.sum(duration, [1, 2]), 1).long()


        # Create a padding mask for the output lengths of shape (batch, 1, max_output_length)
        indices = torch.arange(predicted_lengths.max(), dtype=predicted_lengths.dtype, device=predicted_lengths.device)
        output_padding_mask = indices.unsqueeze(0) < predicted_lengths.unsqueeze(1)
        output_padding_mask = output_padding_mask.unsqueeze(1).to(input_padding_mask.dtype)

        # Reconstruct an attention tensor of shape (batch, 1, out_length, in_length)
        attn_mask = torch.unsqueeze(input_padding_mask, 2) * torch.unsqueeze(output_padding_mask, -1)
        batch_size, _, output_length, input_length = attn_mask.shape
        cum_duration = torch.cumsum(duration, -1).view(batch_size * input_length, 1)
        indices = torch.arange(output_length, dtype=duration.dtype, device=duration.device)
        valid_indices = indices.unsqueeze(0) < cum_duration
        valid_indices = valid_indices.to(attn_mask.dtype).view(batch_size, input_length, output_length)
        padded_indices = valid_indices - nn.functional.pad(valid_indices, [0, 0, 1, 0, 0, 0])[:, :-1]
        attn = padded_indices.unsqueeze(1).transpose(2, 3) * attn_mask

        # Expand prior distribution
        prior_means = torch.matmul(attn.squeeze(1), prior_means).transpose(1, 2)
        prior_log_variances = torch.matmul(attn.squeeze(1), prior_log_variances).transpose(1, 2)

        prior_latents = prior_means + torch.randn_like(prior_means) * torch.exp(prior_log_variances) * self.noise_scale
        latents = self.flow(prior_latents, output_padding_mask, speaker_embeddings, reverse=True)

        spectrogram = latents * output_padding_mask

        for i in range(0, spectrogram.size(-1), chunk_size):
              yield self.decoder(spectrogram[:,:,i : i + chunk_size] ,speaker_embeddings)



api_key = os.environ.get("Id_mode_vits") 
headers = {"Authorization": f"Bearer {api_key}"}

from transformers import AutoTokenizer,VitsModel
import torch
models= {}
tokenizer = AutoTokenizer.from_pretrained("asg2024/vits-ar-sa-huba",token=api_key)
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
def  get_model(name_model):
    global models
    if name_model in   models:
        return models[name_model]
    models[name_model]=VitsModel.from_pretrained(name_model,token=api_key).to(device)
    return models[name_model]

def  genrate_speech(text,name_model):
    inputs=tokenizer(text,return_tensors="pt")
    model=get_model(name_model) 
    with torch.no_grad():
         wav=model(
             input_ids= inputs.input_ids.to(device),
             attention_mask=inputs.attention_mask.to(device),
             speaker_id=0
             ).waveform.cpu().numpy().reshape(-1)
    return model.config.sampling_rate,wav

def generate_audio(text,name_model,speaker_id=None):
    inputs = tokenizer(text, return_tensors="pt")#.input_ids

    speaker_embeddings = None
    model=get_model(name_model)
    #torch.cuda.empty_cache()
    with torch.no_grad():
        for chunk in _inference_forward_stream(model,input_ids=inputs.input_ids,attention_mask=inputs.attention_mask,speaker_embeddings= speaker_embeddings,chunk_size=64):
            yield  16000,chunk.squeeze().cpu().numpy()#.astype(np.int16).tobytes()    

def remove_extra_spaces(text):
  """
  Removes extra spaces between words in a string.

  Args:
    text: The string to process.

  Returns:
    The string with extra spaces removed.
  """
  return ' '.join(text.split())

def query(text,API_URL):
  payload={"inputs": text}
  response = requests.post(API_URL, headers=headers, json=payload)
  return response.content

def   get_answer_ai(text):
      global AI
      try:
          response = AI.send_message(text)
          return response.text

          
      except :
          AI=create_chat_session()
          response = AI.send_message(text)
          return response.text

with gr.Blocks() as demo:  # Use gr.Blocks to wrap the entire interface
    with gr.Column():
                model_choices = gr.Dropdown(
                    choices=[
                        "asg2024/vits-ar-sa",
                        "asg2024/vits-ar-sa-huba",
                        "asg2024/vits-ar-sa-ms",
                        "asg2024/vits-ar-sa-magd",
                        "asg2024/vits-ar-sa-fahd",
                    ],
                    label="اختر النموذج",
                    value="asg2024/vits-ar-sa",
                )
    with gr.Tab("Chat AI "):
        gr.Markdown("## AI: محادثة صوتية بالذكاء الاصطناعي باللهجة السعودية")
        with gr.Row(): # Arrange input/output components side-by-side
            with gr.Column():
                text_input = gr.Textbox(label="أدخل أي نص")
                user_audio = gr.Audio(label="صوتك")
                with gr.Row():
                    btn = gr.Button("إرسال")
                    btn_ai_only = gr.Button("توليد رد الذكاء الاصطناعي فقط")
                    
            
                ai_audio = gr.Audio(label="رد الذكاء الاصطناعي الصوتي")
                ai_text = gr.Textbox(label="رد الذكاء الاصطناعي النصي")

       

        # Use a single button to trigger both functionalities
        def process_audio(text, model_choice, generate_user_audio=True):
            API_URL = f"https://api-inference.huggingface.co/models/{model_choice}"
            text_answer = get_answer_ai(text)
            text_answer = remove_extra_spaces(text_answer)
            data_ai = genrate_speech(text_answer,model_choice)#query(text_answer, API_URL)
            if generate_user_audio:  # Generate user audio if needed
                data_user =genrate_speech(text,model_choice)# query(text, API_URL)
                return data_user, data_ai, text_answer
            else:
                return  data_ai  # Return None for user_audio

        btn.click(
            process_audio,  # Call the combined function
            inputs=[text_input, model_choices],
            outputs=[user_audio, ai_audio, ai_text],
        )

        # Additional button to generate only AI audio
      
        btn_ai_only.click(
                lambda text, model_choice: process_audio(text, model_choice, False),
                inputs=[text_input, model_choices],
                outputs=[ai_audio],
            )
    with gr.Tab("Live "):
        gr.Markdown("## VITS: تحويل النص إلى كلام")
        with gr.Row():
            text_input = gr.Textbox(label="أدخل النص هنا")
            speaker_id_input = gr.Number(label="معرّف المتحدث (اختياري)", interactive=True)
        generate_button = gr.Button("توليد وتشغيل الصوت")
      
        audio_player = gr.Audio(label="أ audio",streaming=True)
    
        # Update the event binding
        generate_button.click(generate_audio, inputs=[text_input,model_choice], outputs=audio_player)

        
        
 
if __name__ == "__main__":
    demo.launch()