File size: 3,630 Bytes
21a655c aff19d1 83f5881 aff19d1 83f5881 aff19d1 a73412a e8f9b13 83f5881 e8f9b13 83f5881 aff19d1 e8f9b13 83f5881 e8f9b13 aff19d1 0abf22c e8f9b13 0abf22c cf00e6b 0abf22c 364fbfd aff19d1 9bab38b 83f5881 aff19d1 83f5881 aff19d1 c406197 364fbfd aff19d1 0cf65dc efb7dae |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 |
#app.py
import gradio as gr
import openai
import os
import RiverValleyData # Importing the RiverValleyData module
import base64
OPENAI_API_KEY = os.getenv("OPENAI_API_KEY")
openai.api_key = OPENAI_API_KEY
def image_to_base64(img_path):
with open(img_path, "rb") as img_file:
return base64.b64encode(img_file.read()).decode('utf-8')
img_base64 = image_to_base64("RiverValleySBC.JPG")
img_html = f'<img src="data:image/jpg;base64,{img_base64}" alt="SBC6" width="300" style="display: block; margin: auto;"/>'
def predict(question_choice, audio):
# Transcribe the audio using Whisper
with open(audio, "rb") as audio_file:
transcript = openai.Audio.transcribe("whisper-1", audio_file)
message = transcript["text"] # This is the transcribed message from the audio input
# Generate the system message based on the chosen question
strategy, explanation = RiverValleyData.strategy_text["SEP"]
# Reference to the picture description from RiverValleyData.py
picture_description = RiverValleyData.description
# Determine whether to include the picture description based on the question choice
picture_description_inclusion = f"""
For the first question, ensure your feedback refers to the picture description provided:
{picture_description}
""" if question_choice == RiverValleyData.questions[0] else ""
# Construct the conversation with the system and user's message
conversation = [
{
"role": "system",
"content": f"""
You are an expert English Language Teacher in a Singapore Primary school, directly guiding a Primary 6 student in Singapore.
The student is answering the question: '{question_choice}'.
{picture_description_inclusion}
Point out areas they did well and where they can improve, following the {strategy}.
Encourage the use of sophisticated vocabulary and expressions.
For the second and third questions, the picture is not relevant, so the student should not refer to it in their response.
{explanation}
The feedback should be in second person, addressing the student directly.
"""
},
{"role": "user", "content": message}
]
response = openai.ChatCompletion.create(
model='gpt-3.5-turbo',
messages=conversation,
temperature=0.6,
max_tokens=500, # Limiting the response to 500 tokens
stream=True
)
partial_message = ""
for chunk in response:
if len(chunk['choices'][0]['delta']) != 0:
partial_message = partial_message + chunk['choices'][0]['delta']['content']
yield partial_message
# Gradio Interface
iface = gr.Interface(
fn=predict,
inputs=[
gr.Radio(RiverValleyData.questions, label="Choose a question", default=RiverValleyData.questions[0]), # Dropdown for question choice
gr.inputs.Audio(source="microphone", type="filepath") # Audio input
],
outputs=gr.inputs.Textbox(), # Using inputs.Textbox as an output to make it editable
description=img_html + '''
<div style="text-align: center; font-size: medium;">
<a href="https://forms.moe.edu.sg/forms/J0lmkJ" target="_blank">
📝 Click here to provide feedback on the initial prototype of the Oral Coach 📝
</a>
</div>
''', # Added feedback link, centralized with medium font size and emoticons
css="custom.css" # Link to the custom CSS file
)
iface.queue(max_size=99, concurrency_count=40).launch(debug=True)
|