Upload 2 files

app.py

@@ -0,0 +1,233 @@
+import openai
+import whisper
+from io import BytesIO
+import os
+import sys
+import pytube
+from moviepy.editor import VideoFileClip
+import moviepy.editor as movpy
+from moviepy.editor import AudioFileClip
+from pydub import AudioSegment
+from pytube import YouTube
+import gradio as gr
+import collections
+from gtts import gTTS
+
+##############################################################################
+openai.api_key = "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"
+#create gradio app
+class NamedBytesIO(BytesIO):
+    def __init__(self, *args, **kwargs):
+        self._name = kwargs.pop('name', 'unnamed.mp3')
+        super(NamedBytesIO, self).__init__(*args, **kwargs)
+
+    @property
+    def name(self):
+        return self._name
+
+##############################################################################       
+def transcribe_audio(audio_file_path):
+    model_id = 'whisper-1'
+
+    # audio_file_path = "C:\VOLUNTEER\TOASTMASTER\Tell me about yourself - bad graduate job interview answer tell me about yourselfout.mp3"
+    audio_file_path = audio_file_path
+    audio = AudioSegment.from_file(audio_file_path, format="mp3")
+
+    audio_duration = int(audio.duration_seconds)
+
+    print(f"Duration of the audio is {audio_duration} seconds")
+    chunk_duration = 60 * 1000 * 20 # 20 minute in milliseconds
+
+    transcriptions = []
+
+    for i in range(0, audio_duration, chunk_duration):
+        chunk = audio[i:i + chunk_duration]
+        chunk_buffer = NamedBytesIO(name='chunk.mp3')
+        chunk.export(chunk_buffer, format="mp3")
+        chunk_buffer.seek(0)
+        
+        response = openai.Audio.transcribe(
+            api_key=openai.api_key,
+            model=model_id,
+            file=chunk_buffer,
+            prompt="""
+            Dont remove any filler words in transcribe. eg: um, umm, uh, ah, er, mhm, hmm, mm, mmm, oh, ohh, let  me think like, Okay, ok, here's what I'm, like, thinking, you know,  well, so, actually, basically, literally, right, i mean,anyway... 
+            This is human voices with various tone and accents. 
+            1) Transcibe every words and all signals, dont skip any  
+            2) Dont correct the grammar
+            3) Dont correct the spelling
+            4) Dont remove any redundant words or punctuations
+            5) if there is a pause in the audio, please add a comma(,) in the transcribe
+            6) if there is a long pause in the audio, please add a period(.) in the transcribe
+            """
+        )
+        
+        transcriptions.append(response['text'])
+
+    final_transcription = ' '.join(transcriptions)
+
+    print(final_transcription.replace('.', '.\n'))
+
+    return final_transcription
+#########################################################################################
+
+def get_feedback(audio_file_path):
+
+    transcribed_text = transcribe_audio(audio_file_path)
+    # response = final_transcription
+
+    prompt = f"""
+    Pretend that you are an interview coach with 25 years of experience.
+    Evaluate the following response based on the evaluation criteria:
+    # please provide feedback on the following response based on the evaluation criteria: 
+    # language, tone & personality, and a 5-point system for clarity, vocal variety, comfort level, interest, and well-supported content.
+
+    Evaluation criteria:
+    A) Language:
+        1. use of Filler words: count the filler words and list them: eg: um, umm, uh, ah, er, mhm, hmm, mm, mmm, oh, ohh, let  me think like, Okay, ok, here's what I'm, like, thinking, you know,  well, so, actually, basically, literally, right, i mean,anyway... 
+            example: um: 3, yeah: 2
+        2. use of redundant words: list them: example: sum total, joint collaboration, unexpected surprise, future plans, new record...etc
+            example: sum total: 2, joint collaboration: 1
+        3. use of jargons: list them
+
+    B) Tone & personality:
+        1. Tone: What is the Tone?
+                A. Positive/optimistic: Expressing a hopeful or upbeat outlook, such as excitement, joy, or satisfaction. Example: "I'm really looking forward to this vacation!"
+                B. Negative/pessimistic: Expressing a negative or gloomy outlook, such as sadness, frustration, or disappointment. Example: "I don't think I can handle any more bad news."
+                C. Neutral/objective: Expressing an unbiased or factual outlook, without any emotion or bias. Example: "The temperature is 75 degrees and the sky is clear."
+                D. Sarcastic: Expressing a tone of mockery or irony, often with the opposite meaning of what is being said. Example: "Oh great, another Monday morning. Just what I needed."
+                E. Formal/polite: Using formal language and expressions to show respect and politeness. Example: "I would be most grateful if you could provide me with further information."
+                F. Informal/casual: Using casual language and expressions to show familiarity and informality. Example: "Hey, what's up? Wanna hang out later?"
+                G. Authoritative: Expressing a tone of authority or control, such as in instructions or commands. Example: "You need to follow these procedures precisely to ensure safety."
+                H. Condescending: Expressing a tone of superiority or patronization towards others.
+        2. Personality traits: Insights into the personality traits?
+                A. Extraversion: Extraverted individuals tend to be outgoing, sociable, and talkative.
+                B. Introversion: Introverted individuals tend to be more reserved, reflective, and introspective. 
+                C. Conscientiousness: Conscientious individuals tend to be organized, responsible, and diligent.
+                D. Agreeableness: Agreeable individuals tend to be friendly, cooperative, and empathetic.
+                E. Neuroticism: Neurotic individuals tend to be anxious, sensitive, and easily stressed.
+        3. Relevance: Does the response address the question asked and provide relevant information?
+        4. Clarity: Is the response clear and concise?
+        5. Specificity / examples or details: Does the response provide specific examples or details that demonstrate the interviewee's skills, experiences, or qualifications?
+        6. Confidence: Does the interviewee present themselves confidently and effectively communicate their ideas?
+        7. Fit: Does the response demonstrate how the interviewee's skills, experiences, or qualifications align with the job requirement?
+
+    C) 3 point system: The user to be evaluated on the following criteria with a 3 point system: 
+
+        1. Clarity: Spoken language is clear and is easily understood Comment 
+
+            3 Is an exemplary Interviewee who is always understood. 
+            2 Spoken language is clear and is easily understood 
+            1 Spoken language is unclear or not easily understood 
+
+        2. Comfort Level 
+
+            3 Appears completely self-assured with the Interviewer
+            2 Appears comfortable with the Interviewer 
+            1 Appears highly uncomfortable with the Interviewer 
+
+        3. Interesting content: Engages Interviewer with interesting, well-constructed content Comment 
+
+            3 Fully engages Interviewer with exemplary, well constructed content 
+            2 Engages Interviewer with interesting, well constructed content 
+            1 Content is neither interesting nor well-constructed 
+
+        4. Well Supported: Speech content is well-supported and sources are available if requested 
+
+            3 Delivers exemplary speech with a topic that is well-supported by content of the speech 
+            2 Speech topic is well-supported by content of speech 
+            1 Speech content is unrelated to the topic of the speech
+
+    Assess the response and give your honest feedback. 
+    A) Language 
+    B) Tone & personality
+    C) 3 point evaluation: Overall Score = sum of points scored "/" 12
+
+    Put in top 3 bullet points for feedback + 3 bullet points on how to improve it. 
+
+    Response:
+    {transcribed_text}
+    """
+
+    # Call the OpenAI API
+    api_response = openai.Completion.create(
+        engine="text-davinci-003",  # Use "text-davinci-002" for GPT-3.5, replace with the appropriate engine name for GPT-4 if available
+        prompt=prompt,
+        temperature=0,
+        max_tokens=500,
+        top_p=1.0,   
+        frequency_penalty=0.0,
+        presence_penalty=0.0
+    )
+
+    # Extract and return the generated feedback
+    feedback = api_response.choices[0].text.strip()
+       
+    language = "en"
+    audioobj = gTTS(text = feedback, 
+                    lang = language, 
+                    slow = False)
+    
+    audioobj.save("Temp.mp3")
+
+    return [feedback, transcribed_text, "Temp.mp3"]
+
+
+
+# iface = gr.Interface(
+#     fn=get_feedback,
+#     inputs=gr.inputs.Textbox(lines=10, label="Interviewee audio file path"),
+#     outputs=[
+#         gr.outputs.Textbox(label="Feedback"),
+#         gr.outputs.Textbox(label="Transcription")
+#     ],
+#     title="Interview Feedback",
+#     examples=[
+#         "Tell me about yourself - bad graduate job interview answer tell me about yourselfout.mp3",
+#         "Tell me about yourself - bad entrepreneur answer to tell me about yourselfout.mp3"
+#     ],
+#     description="Get feedback on your interview response from an AI interview coach.",
+#     layout="vertical"
+# )
+
+# iface.launch()
+
+
+# Create a Gradio interface
+# the layout should be input, examples, feedback, transcription
+# the input should be a textbox
+# the examples should be a list of audio files
+# the output should be a textbox for feedback and a textbox for transcription
+
+iface = gr.Interface(
+    fn=get_feedback,
+    inputs=gr.inputs.Textbox(lines=10, label="Interviewee audio file path"),
+    outputs=[
+        gr.outputs.Textbox(label="Feedback"),
+        gr.outputs.Textbox(label="Transcription"),
+        gr.Audio("Temp.mp3", label="Speech Output")
+    ],
+    title="Interview Feedback",
+    examples=[
+        "Tell me about yourself - bad graduate job interview answer tell me about yourselfout.mp3",
+        "Tell me about yourself - bad entrepreneur answer to tell me about yourselfout.mp3"
+    ],
+    description="Get feedback on your interview response from an AI interview coach.",
+    layout="vertical"
+)
+iface.launch()
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+#########################################################################################

requirements.txt

@@ -0,0 +1,7 @@
+openai
+whisper
+pytube
+moviepy
+pydub
+gradio
+gtts