Update app.py

app.py

@@ -13,6 +13,12 @@ import requests
 from bs4 import BeautifulSoup
 import urllib
 import random
+from functools import lru_cache
+import concurrent.futures
+
+# Configuration for concurrency
+MAX_WORKERS = 4  # Adjust based on your system resources
+executor = concurrent.futures.ThreadPoolExecutor(max_workers=MAX_WORKERS)
 
 # List of user agents to choose from for requests
 _useragent_list = [
@@ -89,15 +95,20 @@ def search(term, num_results=3, lang="en", advanced=True, timeout=5, safe="activ
             start += len(result_block)
     return all_results
 
+@lru_cache(maxsize=1)  # Cache the models to avoid reloading
+def load_speech_recognition_models():
+    """Loads and caches speech recognition models."""
+    model_name = "neongeckocom/stt_en_citrinet_512_gamma_0_25"
+    sample_rate = 16000
+    preprocessor = torch.jit.load(hf_hub_download(model_name, "preprocessor.ts", subfolder="onnx"))
+    encoder = ort.InferenceSession(hf_hub_download(model_name, "model.onnx", subfolder="onnx"))
+    tokenizer = spm.SentencePieceProcessor(hf_hub_download(model_name, "tokenizer.spm", subfolder="onnx"))
+    return preprocessor, encoder, tokenizer
+
 # Speech Recognition Model Configuration
 model_name = "neongeckocom/stt_en_citrinet_512_gamma_0_25"
 sample_rate = 16000
 
-# Download preprocessor, encoder and tokenizer
-preprocessor = torch.jit.load(hf_hub_download(model_name, "preprocessor.ts", subfolder="onnx"))
-encoder = ort.InferenceSession(hf_hub_download(model_name, "model.onnx", subfolder="onnx"))
-tokenizer = spm.SentencePieceProcessor(hf_hub_download(model_name, "tokenizer.spm", subfolder="onnx"))
-
 # Mistral Model Configuration
 client1 = InferenceClient("mistralai/Mixtral-8x7B-Instruct-v0.1")
 system_instructions1 = "<s>[SYSTEM] Answer as Real OpenGPT 4o, Made by 'KingNish', Keep conversation very short, clear, friendly and concise. The text provided is a request for a specific type of response from you, the virtual assistant. The request asks you to provide friendly responses. The expectation is that I will avoid introductions and start answering the query directly, elaborating on all aspects of the request. As an AI-powered assistant, your task is to generate responses that appear as if they are created by the user and not an AI-powered assistant. [USER]"
@@ -109,6 +120,8 @@ def to_float32(audio_buffer):
     return np.divide(audio_buffer, np.iinfo(audio_buffer.dtype).max, dtype=np.float32)
 
 def transcribe(audio_path):
+    """Transcribes audio using cached models."""
+    preprocessor, encoder, tokenizer = load_speech_recognition_models()
     audio_file = AudioSegment.from_file(audio_path)
     sr = audio_file.frame_rate
     audio_buffer = np.array(audio_file.get_array_of_samples())
@@ -128,34 +141,38 @@ def transcribe(audio_path):
 
     return text
 
-def model(text, web_search):
-    if web_search is True:
-        """Performs a web search, feeds the results to a language model, and returns the answer."""
-        web_results = search(text)
+async def run_model(text, web_search):
+    """Runs the language model asynchronously."""
+    if web_search:
+        web_results = await asyncio.get_event_loop().run_in_executor(executor, search, text)  # Run search in executor
         web2 = ' '.join([f"Link: {res['link']}\nText: {res['text']}\n\n" for res in web_results])
         formatted_prompt = system_instructions1 + text + "[WEB]" + str(web2) + "[ANSWER]"
-        stream = client1.text_generation(formatted_prompt, max_new_tokens=512, stream=True, details=True, return_full_text=False)
-        return "".join([response.token.text for response in stream if response.token.text != "</s>"])
     else:
         formatted_prompt = system_instructions1 + text + "[JARVIS]"
-        stream = client1.text_generation(formatted_prompt, max_new_tokens=512, stream=True, details=True, return_full_text=False)
-        return "".join([response.token.text for response in stream if response.token.text != "</s>"])
+    stream = client1.text_generation(formatted_prompt, max_new_tokens=512, stream=True, details=True, return_full_text=False)
+    return "".join([response.token.text for response in stream if response.token.text != "</s>"])
 
-async def respond(audio, web_search):
-    user = transcribe(audio)
-    reply = model(user, web_search)
+async def generate_speech(reply):
+    """Generates speech asynchronously."""
     communicate = edge_tts.Communicate(reply)
     with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as tmp_file:
         tmp_path = tmp_file.name
         await communicate.save(tmp_path)
     return tmp_path
 
+async def respond(audio, web_search):
+    """Handles user input, model processing, and response generation."""
+    user = await asyncio.get_event_loop().run_in_executor(executor, transcribe, audio)  # Run transcription in executor
+    reply = await run_model(user, web_search)
+    audio_path = await generate_speech(reply)
+    return audio_path
+
 with gr.Blocks() as demo:    
     with gr.Row():
         web_search = gr.Checkbox(label="Web Search", value=False)
-        input = gr.Audio(label="Voice Chat", sources="microphone")
+        input = gr.Audio(label="Voice Chat", sources="microphone", type="numpy")
         output = gr.Audio(label="AI",autoplay=True)
-        gr.Interface(fn=respond, inputs=[input, web_search], outputs=[output], live=True, batch=True, max_batch_size=20, delete_cache=(60,60))
+        gr.Interface(fn=respond, inputs=[input, web_search], outputs=[output], live=True)
  
 if __name__ == "__main__":
     demo.queue(max_size=200).launch()