first commit

app.py

@@ -0,0 +1,66 @@
+import os
+import streamlit as st
+from audiorecorder import audiorecorder
+from apiclient import discovery
+from httplib2 import Http
+from oauth2client import client, file, tools
+
+
+import warnings
+
+from transcription import transcribe
+from quiz_generation import generate_quiz_url, explain_quiz_answers
+
+SCOPES = "https://www.googleapis.com/auth/forms.body"
+
+
+def main():
+    warnings.filterwarnings("ignore")
+
+    # Initialize Google Sheets and Forms API services
+    store = file.Storage("credentials.json")
+    creds = store.get()
+    if not creds or creds.invalid:
+        flow = client.flow_from_clientsecrets(
+            r"C:\Users\Admin\Downloads\client_secret_535279977482-ttq1qb18v1crma5bkf70015qk9e9r2vv.apps.googleusercontent.com.json",
+            SCOPES
+        )
+        creds = tools.run_flow(flow, store)
+    form_service = discovery.build("forms", "v1", http=creds.authorize(Http()))
+
+    st.title("Quiz Generator")
+    st.markdown("Record an audio clip and generate a quiz based on the transcribed text.")
+    audio = audiorecorder("Click to record", "Stop recording")
+
+    if len(audio) > 0:
+        # To play audio in the frontend:
+        st.audio(audio.tobytes(), format="audio/wav")
+
+        # To save audio to a file:
+        wav_file = open("audio.wav", "wb")
+        wav_file.write(audio.tobytes())
+
+    # Quiz generation section
+    st.header("Quiz Generation")
+
+    if st.button("Generate Quiz"):
+        with st.spinner("Transcribing audio to generate the quiz..."):
+            #transcribed_text = transcribe("audio.wav")
+            transcribed_text = " can you please generate a quiz of 4 questions about ML, each of them with 4 answers and indicate the correct answer"
+            # Get the explanations for the quiz
+            
+            
+            quiz_url, explanations = generate_quiz_url(transcribed_text, form_service)
+            st.success("Quiz generated successfully!")
+            st.text("Quiz Link: " + quiz_url)
+            st.text("Transcribed Text:\n" + transcribed_text)
+            
+            # Display the explanations
+            st.header("Quiz Explanations")
+            for i, explanation in enumerate(explanations):
+                st.subheader(f"Question {i+1}")
+                st.text(explanation)
+
+
+if __name__ == '__main__':
+    main()

credentials.json

@@ -0,0 +1 @@
+{"access_token": "ya29.a0AWY7Cknnaz0R8i2DlngKKKx4C_IZKzUVFPdZqk-e7diB_cGu1FcQDncMZArWOrjejGLUHakobPGddDruqRMC5Eu5ZKopv4BsKFPJi9mDLEwJBh8a7cYuIjMTZIQMkHQMtDm1Oz9T-QWjf26tLo_3iKKMOX7Gds8aCgYKARYSARESFQG1tDrpuJfmVxN7kf1ZQkwiDIKA5g0166", "client_id": "535279977482-ttq1qb18v1crma5bkf70015qk9e9r2vv.apps.googleusercontent.com", "client_secret": "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", "refresh_token": "1//03cifxY_-1uh0CgYIARAAGAMSNwF-L9IrA86QsxrDPYOR3JWrekwFt42ZYG5RCssKeYOv0YWqEwEr75FCT6S5hEloEG2wKomo91c", "token_expiry": "2023-05-27T18:43:18Z", "token_uri": "https://oauth2.googleapis.com/token", "user_agent": null, "revoke_uri": "https://oauth2.googleapis.com/revoke", "id_token": null, "id_token_jwt": null, "token_response": {"access_token": "ya29.a0AWY7Cknnaz0R8i2DlngKKKx4C_IZKzUVFPdZqk-e7diB_cGu1FcQDncMZArWOrjejGLUHakobPGddDruqRMC5Eu5ZKopv4BsKFPJi9mDLEwJBh8a7cYuIjMTZIQMkHQMtDm1Oz9T-QWjf26tLo_3iKKMOX7Gds8aCgYKARYSARESFQG1tDrpuJfmVxN7kf1ZQkwiDIKA5g0166", "expires_in": 3599, "scope": "https://www.googleapis.com/auth/forms.body", "token_type": "Bearer"}, "scopes": ["https://www.googleapis.com/auth/forms.body"], "token_info_uri": "https://oauth2.googleapis.com/tokeninfo", "invalid": false, "_class": "OAuth2Credentials", "_module": "oauth2client.client"}

quiz_generation.py

@@ -0,0 +1,227 @@
+import re
+import os
+from apiclient import discovery
+from oauth2client import client, file, tools
+import bardapi
+from transformers import pipeline, AutoModelForQuestionAnswering, AutoTokenizer
+from transformers import T5ForConditionalGeneration, T5Tokenizer
+
+SCOPES = "https://www.googleapis.com/auth/forms.body"
+DISCOVERY_DOC = "https://forms.googleapis.com/$discovery/rest?version=v1"
+
+NEW_FORM = {
+    "info": {
+        "title": "Quiz"
+    }
+}
+
+model_name = "t5-base"
+model = T5ForConditionalGeneration.from_pretrained(model_name)
+tokenizer = T5Tokenizer.from_pretrained(model_name)
+
+def generate_quiz_questions(prompt):
+    # Set your Bard API key as an environment variable
+    os.environ['_BARD_API_KEY'] = "WwgqSrcbBC71HsiWpTlqnbDC9TQ3-9N1YyY6CHxOEfFp_qeCe0laziZoOT_dkTEjhJmOcw."
+    
+    prompt_suffix = ". Each generated question has to begin with '🔹', each choice has to begin with '🔸', and each correct answer has to begin with '✔️'."
+
+
+    # Send API requests and get responses
+    response = bardapi.core.Bard().get_answer(prompt + prompt_suffix)
+
+    quiz = response["content"]
+
+    return quiz
+
+'''
+def generate_quiz_url(prompt_text, form_service):
+    # Generate quiz questions based on the transcribed text
+    text = generate_quiz_questions(prompt_text)
+
+    # Questions, choices, and correct answers
+    questions = re.findall(r"🔹 (.*?)\n", text)
+    choices = re.findall(r"🔸 (.*?)\n", text)
+    answers = re.findall(r"✔️ (.*?)\n", text)
+
+    # Remove the '**' from the questions list (they are not part of the question), choices, and correct answers
+    questions = [question.replace('**', '') for question in questions]
+    answers = [answer.replace('**', '') for answer in answers]
+
+    questions_list = []
+
+    # Fill the questions_list variable
+    for i, question in enumerate(questions):
+        choices_for_question = choices[i * 4:(i + 1) * 4]
+        correct_answer = answers[i] if i < len(answers) else ""
+
+        question_dict = {
+            "question": question,
+            "choices": choices_for_question,
+            "correct_answer": correct_answer
+        }
+
+        questions_list.append(question_dict)
+
+    # Create the initial form
+    result = form_service.forms().create(body=NEW_FORM).execute()
+
+    # Add the questions to the form
+    question_requests = []
+    for index, question in enumerate(questions_list):
+        new_question = {
+            "createItem": {
+                "item": {
+                    "title": question["question"],
+                    "questionItem": {
+                        "question": {
+                            "required": True,
+                            "choiceQuestion": {
+                                "type": "RADIO",
+                                "options": [
+                                    {"value": choice} for choice in question["choices"]
+                                ],
+                                "shuffle": True
+                            }
+                        }
+                    }
+                },
+                "location": {
+                    "index": index
+                }
+            }
+        }
+        question_requests.append(new_question)
+
+    NEW_QUESTIONS = {
+        "requests": question_requests
+    }
+
+    question_setting = form_service.forms().batchUpdate(formId=result["formId"], body=NEW_QUESTIONS).execute()
+
+    # Retrieve the updated form result
+    get_result = form_service.forms().get(formId=result["formId"]).execute()
+
+    # Get the form ID
+    form_id = get_result["formId"]
+
+    # Construct the quiz link using the form ID
+    form_url = result["responderUri"]
+
+    return form_url
+'''
+
+def explain_quiz_answers(questions_list):
+    explanations = []
+
+    for question in questions_list:
+        context = question["question"]
+        choices = question["choices"]
+        correct_answer = question["correct_answer"]
+
+        explanation = f"Question: {context}\n"
+
+        for choice in choices:
+            # Construct a query with each choice as a question
+            query = f"What is the reason for choosing '{choice}' in {context}?"
+
+            # Tokenize the query and context
+            inputs = tokenizer.encode_plus(query, context, return_tensors="pt", truncation=True, padding="max_length", max_length=512)
+
+            # Generate the explanation using the T5 model
+            outputs = model.generate(input_ids=inputs["input_ids"], attention_mask=inputs["attention_mask"], max_length=256)
+
+            # Decode the explanation
+            explanation_text = tokenizer.decode(outputs[0], skip_special_tokens=True)
+
+            # Add the explanation to the overall explanation
+            explanation += f"\nChoice: {choice}\nExplanation: {explanation_text}"
+
+            # Add an indicator if the choice is the correct answer
+            if choice == correct_answer:
+                explanation += " (Correct Answer)"
+
+            explanation += "\n"
+
+        explanations.append(explanation)
+
+    return explanations
+
+
+def generate_quiz_url(prompt_text, form_service):
+    # Generate quiz questions based on the transcribed text
+    text = generate_quiz_questions(prompt_text)
+
+    # Questions, choices, and correct answers
+    questions = re.findall(r"🔹 (.*?)\n", text)
+    choices = re.findall(r"🔸 (.*?)\n", text)
+    answers = re.findall(r"✔️ (.*?)\n", text)
+
+    # Remove the '**' from the questions list (they are not part of the question), choices, and correct answers
+    questions = [question.replace('**', '') for question in questions]
+    answers = [answer.replace('**', '') for answer in answers]
+
+    questions_list = []
+
+    # Fill the questions_list variable
+    for i, question in enumerate(questions):
+        choices_for_question = choices[i * 4:(i + 1) * 4]
+        correct_answer = answers[i] if i < len(answers) else ""
+
+        question_dict = {
+            "question": question,
+            "choices": choices_for_question,
+            "correct_answer": correct_answer
+        }
+
+        questions_list.append(question_dict)
+
+    # Create the initial form
+    result = form_service.forms().create(body=NEW_FORM).execute()
+
+    # Add the questions to the form
+    question_requests = []
+    for index, question in enumerate(questions_list):
+        new_question = {
+            "createItem": {
+                "item": {
+                    "title": question["question"],
+                    "questionItem": {
+                        "question": {
+                            "required": True,
+                            "choiceQuestion": {
+                                "type": "RADIO",
+                                "options": [
+                                    {"value": choice} for choice in question["choices"]
+                                ],
+                                "shuffle": True
+                            }
+                        }
+                    }
+                },
+                "location": {
+                    "index": index
+                }
+            }
+        }
+        question_requests.append(new_question)
+
+    NEW_QUESTIONS = {
+        "requests": question_requests
+    }
+
+    question_setting = form_service.forms().batchUpdate(formId=result["formId"], body=NEW_QUESTIONS).execute()
+
+    # Retrieve the updated form result
+    get_result = form_service.forms().get(formId=result["formId"]).execute()
+
+    # Get the form ID
+    form_id = get_result["formId"]
+
+    # Construct the quiz link using the form ID
+    form_url = result["responderUri"]
+
+    # Get the explanations for the quiz
+    explanations = explain_quiz_answers(questions_list)
+
+    return form_url, explanations
+

transcription.py

@@ -0,0 +1,53 @@
+from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
+import torch
+import whisper
+
+
+
+tokenizer = AutoTokenizer.from_pretrained("Bhuvana/t5-base-spellchecker")
+
+model = AutoModelForSeq2SeqLM.from_pretrained("Bhuvana/t5-base-spellchecker")
+
+
+def correct(inputs):
+    input_ids = tokenizer.encode(inputs,return_tensors='pt')
+    sample_output = model.generate(
+        input_ids,
+        do_sample=True,
+        max_length=50,
+        top_p=0.99,
+        num_return_sequences=1
+    )
+    res = tokenizer.decode(sample_output[0], skip_special_tokens=True)
+    return res
+
+whisper_model = whisper.load_model("base")
+def transcribe(audio_file):
+    # Load audio and pad/trim it to fit 30 seconds
+    audio = whisper.load_audio(audio_file)
+    audio = whisper.pad_or_trim(audio)
+
+    # Convert audio data to PyTorch tensor and float data type
+    mel = torch.from_numpy(audio).float()
+
+    # Make log-Mel spectrogram and move to the same device as the model
+    mel = whisper.log_mel_spectrogram(mel).to(model.device)
+
+    # Detect the spoken language
+    _, probs = whisper_model.detect_language(mel)
+
+    # Decode the audio
+    options = whisper.DecodingOptions(fp16=False)
+    result = whisper.decode(whisper_model, mel, options)
+    result_text = result.text
+    
+    print('result_text:'+result_text)
+
+    return correct(result_text)
+
+
+
+
+
+
+