Add application main

app.py

@@ -1,20 +1,31 @@
+from huggingface_hub import snapshot_download
+
+# Download models from Hugging Face to local folders
+snapshot_download(
+    repo_id="OpenVINO/Mistral-7B-Instruct-v0.2-int4-ov",
+    local_dir="mistral-ov"
+)
+snapshot_download(
+    repo_id="OpenVINO/whisper-tiny-fp16-ov",
+    local_dir="whisper-ov-model"
+)
+
 import gradio as gr
 import openvino_genai
 import librosa
 import numpy as np
-from threading import Thread, Lock,Event
+from threading import Thread, Lock, Event
 from scipy.ndimage import uniform_filter1d
 from queue import Queue, Empty
 
-
 # Initialize Mistral pipeline
 mistral_pipe = openvino_genai.LLMPipeline("mistral-ov", device="CPU")
 config = openvino_genai.GenerationConfig(
-      max_new_tokens=100,  # Increased for better context
-    num_beams=1,         # Keep greedy search for speed
-    do_sample=False,     # Disable sampling for faster generation
-    temperature=0.0,     # Set to 0 since sampling is disabled
-    top_p=1.0,           # Disable top-p filtering
+    max_new_tokens=100,
+    num_beams=1,
+    do_sample=False,
+    temperature=0.0,
+    top_p=1.0,
     top_k=50  
 )
 pipe_lock = Lock()
@@ -54,7 +65,6 @@ def transcribe(audio):
     return whisper_pipe.generate(processed)
 
 def stream_generator(message, history):
-    """Original Mistral streaming function (unchanged)"""
     response_queue = Queue()
     completion_event = Event()
     error_message = [None]
@@ -66,9 +76,7 @@ def stream_generator(message, history):
     def generate():
         try:
             with pipe_lock:
-              
                 mistral_pipe.generate(message, config, callback)
-            
         except Exception as e:
             error_message[0] = str(e)
         finally:
@@ -91,9 +99,7 @@ def stream_generator(message, history):
 
     yield "".join(accumulated)
 
-# Create interface with added voice input
 with gr.Blocks() as demo:
-    # Original chat interface
     chat_interface = gr.ChatInterface(
         stream_generator,
         textbox=gr.Textbox(placeholder="Ask Mistral...", container=False),
@@ -106,12 +112,10 @@ with gr.Blocks() as demo:
         cache_examples=False,
     )
     
-    # Add voice input below examples
     with gr.Row():
         audio = gr.Audio(sources=["microphone"], type="numpy", label="Voice Input")
         transcribe_btn = gr.Button("Send Transcription")
     
-    # Connect transcription to chat input
     transcribe_btn.click(
         transcribe,
         inputs=audio,
@@ -119,4 +123,4 @@ with gr.Blocks() as demo:
     )
 
 if __name__ == "__main__":
-    demo.launch(share=True, debug=True)
+    demo.launch(share=True,debug=True)

requirements.txt

@@ -1,9 +1,6 @@
-gradio>=4.13.0
-openvino-genai>=2024.1.0  # Correct package name
+gradio==4.26.0
+openvino-genai>=1.0.0
 librosa>=0.10.0
-numpy>=1.21.0
+numpy>=1.24.0
 scipy>=1.10.0
-huggingface_hub>=0.19.0
-soundfile>=0.12.0
-torch>=2.2.0  # Required for some components
-transformers>=4.38.0  # Needed for audio processing
+huggingface_hub>=0.21.4