Upload 2 files

app.py

@@ -9,6 +9,11 @@ from langchain_openai import OpenAIEmbeddings
 from langchain_openai import ChatOpenAI
 from langchain.schema import StrOutputParser
 from langchain.prompts import PromptTemplate
+from openai import OpenAI
+from transformers import pipeline
+import numpy as np
+
+
 
 
 
@@ -18,7 +23,7 @@ secret_string= os.getenv('OPENAI_API_KEY')
 #create simple langchain QA from question of user
 
 engine = "ft:gpt-3.5-turbo-1106:sinensia:movies:9PDh7plA"
-
+client = OpenAI(api_key=secret_string)
 llm = ChatOpenAI(model=engine, temperature=0, openai_api_key=secret_string)
 
 prompt = PromptTemplate.from_template(
@@ -36,13 +41,24 @@ answer:
 
 chain = prompt | llm | StrOutputParser()
 
-#make a sinmple gradio interface to interact with the model
-import gradio as gr
+transcriber = pipeline("automatic-speech-recognition", model="openai/whisper-base.en")
+
+#make a sinmple gradio interface that can input text or voice and output text
 
-def ask_question(question):
+def voice_to_text(audio):
+    sr, y = audio
+    y = y.astype(np.float32)
+    y /= np.max(np.abs(y))
+
+    question = transcriber({"sampling_rate": sr, "raw": y})["text"]
+    return answer_question(question)
+
+def answer_question(question):
     return chain.invoke({"input": question})
 
-iface = gr.Interface(fn=ask_question, inputs="text", outputs="text")
+# interface with two inputs, one for text and one for voice and one output for text
+import gradio as gr
 
-iface.launch()
+iface = gr.Interface(fn=lambda text, audio: answer_question(text) if text else voice_to_text(audio), inputs=[gr.Textbox(lines=2, placeholder="Ask a question about movies"), gr.Audio()], outputs="text")
 
+iface.launch()

requirements.txt

@@ -31,4 +31,5 @@ PyPika==0.48.9
 pyproject_hooks==1.1.0
 python-dotenv==1.0.1
 python-multipart==0.0.9
+tf-keras