Create app.py

app.py

@@ -0,0 +1,43 @@
+from transformers import pipeline
+from langchain.vectorstores import FAISS
+from langchain.embeddings import HuggingFaceEmbeddings
+
+pipe = pipeline("automatic-speech-recognition", model="openai/whisper-large-v2")
+import json
+with open("tasks.json", "r",encoding="utf-8") as json_file:
+  global data
+  data = json.load(json_file)
+def find_index(sentence):
+  global data
+  for key, value in data.items():
+    for i,j in value.items():
+      for s in j:
+        if sentence == s:
+          return i
+for x,item in data.items():
+    texts = []
+    for key,value in item.items():
+      for each  in value:
+        print(find_index(each))
+        texts.append(each)
+    globals()[f"faiss_{x}"] = FAISS.from_texts(texts,HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2",model_kwargs={'device':'cpu'}))
+
+from fastapi import FastAPI
+app = FastAPI()
+@app.get("/")
+def transcribe_the_command(audio_path,state):
+    transcript = pipe(audio_path)["text"]
+    similar = globals()[f"faiss_{state}"].similarity_search(transcript)[0].page_content
+    print(similar)
+    reply = find_index(similar)
+    return reply
+import gradio as gr
+iface = gr.Interface(
+    fn=transcribe_the_command,
+    inputs=[gr.Textbox(),gr.Textbox()],
+    outputs="text",
+    title="Whisper Small",
+    description="Realtime demo for intent recognition using a Whisper small model.",
+)
+
+iface.launch(share="true")