yea

app.py

@@ -1,15 +1,16 @@
 import datetime
 import json
-import os
-import polars as pl
+import time
+
 import chromadb
 import gradio as gr
+import polars as pl
 import spaces
 from chromadb.utils import embedding_functions
 from dotenv import load_dotenv
 from langchain.docstore.document import Document
 from langchain.text_splitter import RecursiveCharacterTextSplitter
-from smolagents import Tool, CodeAgent, TransformersModel, stream_to_gradio, tool, ToolCallingAgent, HfApiModel
+from smolagents import Tool, CodeAgent, stream_to_gradio, tool, ToolCallingAgent, HfApiModel
 from sqlalchemy import (
     create_engine,
     MetaData,
@@ -20,15 +21,14 @@ from sqlalchemy import (
     insert,
     text, Numeric, DateTime, func
 )
-from huggingface_hub import login
-from Constants import BAKERY_ORDERS_DATA, BAKING_RECIPES, RAG_QUESTION, SQL_QUERY
 from transformers import pipeline
-import time
+
+from Constants import BAKERY_ORDERS_DATA, BAKING_RECIPES, RAG_QUESTION, SQL_QUERY
+
 load_dotenv()
 # os.system("python -m phoenix.server.main serve")
 import numpy as np
 import os
-import base64
 
 
 # # Get your own keys from https://cloud.langfuse.com
@@ -278,26 +278,10 @@ if __name__ == "__main__":
         return "", []
 
 
-    transcriber = pipeline("automatic-speech-recognition", model="openai/whisper-large-v3-turbo",device_map = 'cuda')
-
-
-    #
-    # def transcribe(audio):
-    #     sr, y = audio
-    #
-    #     # Convert to mono if stereo
-    #     if y.ndim > 1:
-    #         y = y.mean(axis=1)
-    #
-    #     y = y.astype(np.float32)
-    #     y /= np.max(np.abs(y))
-    #
-    #     return transcriber({"sampling_rate": sr, "raw": y})["text"]
     @spaces.GPU
-    def transcribe(audio_chunk):
+    def transcribe(audio_chunk, message):
         if audio_chunk is None:
             return ""
-        print('audio_chunk',audio_chunk)
         sr, y = audio_chunk
 
         # Convert to mono if stereo
@@ -308,28 +292,18 @@ if __name__ == "__main__":
         y /= np.max(np.abs(y))
 
         time.sleep(2)
-        text2=transcriber({"sampling_rate": sr, "raw": y})["text"]
-        return text2
-        # sr, y = audio_chunk
-        # # Convert to mono if stereo
-        # if y.ndim > 1:
-        #     y = y.mean(axis=1)
-        #
-        # y = y.astype(np.float32)
-        # y /= np.max(np.abs(y))
-        #
-        # if stream is not None or stream.shape != []:
-        #     stream = np.concatenate([stream, y])
-        # else:
-        #     stream = y
-        # return stream, transcriber({"sampling_rate": sr, "raw": stream})["text"]
 
+        message = message + " " + transcriber({"sampling_rate": sr, "raw": y})["text"]
+        return message
+
+
+    transcriber = pipeline("automatic-speech-recognition", model="openai/whisper-large-v3-turbo", device_map='cuda')
 
     with gr.Blocks() as b:
         # GUI
         gr.Markdown("# Bakery shope ordering llm multi agent system")
         with gr.Accordion('''
-        open for more description of this space ''',open=False):
+        open for more description of this space ''', open=False):
             gr.Markdown('''
             with self correcting text2sql agent for orders and RAG agent for recipes.
             using smolagents, gradio, HF Spaces, sqlalchemy,langchain for sematic search, chromadb\n
@@ -342,7 +316,8 @@ if __name__ == "__main__":
         with gr.Accordion("Bakery orders data", open=False):
             gr.DataFrame(pl.DataFrame(BAKERY_ORDERS_DATA))
         with gr.Accordion("Baking recipes data", open=False):
-            gr.DataFrame(pl.DataFrame(BAKING_RECIPES,schema=['baking recipes']))
+            gr.DataFrame(pl.DataFrame(BAKING_RECIPES, schema=['baking recipes']))
+
         chatbot = gr.Chatbot(type="messages", height=900)
         message_box = gr.Textbox(lines=1, label="chat message:")
         with gr.Row():
@@ -363,7 +338,9 @@ if __name__ == "__main__":
         mbox_submit_event = message_box.submit(enter_message,
                                                [message_box, chatbot],
                                                [message_box, chatbot])
-        audio_stream = audio_interface.change(transcribe, inputs=[audio_interface],
+
+        audio_stream = audio_interface.change(transcribe, inputs=[audio_interface, message_box
+                                                                  ],
                                               outputs=[message_box
                                                        ])
         rag_q_click_event = rag_q_button.click(enter_message,
@@ -383,4 +360,4 @@ if __name__ == "__main__":
             cancels=[reply_button_click_event, rag_q_click_event, sql_q_click_event, combi_click_event,
                      mbox_submit_event, audio_stream])
 
-    b.launch()
+b.launch()

employment.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:adc3e8999af8e4c83332f62a68b949b75135e4a3c0eb482722d306416bae45f5
+size 478921