Spaces:
Sleeping
Sleeping
Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,123 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
import numpy as np
|
3 |
+
from transformers import pipeline
|
4 |
+
from sentence_transformers import SentenceTransformer
|
5 |
+
from sklearn.metrics.pairwise import cosine_similarity
|
6 |
+
import PyPDF2
|
7 |
+
import torch
|
8 |
+
import gc
|
9 |
+
|
10 |
+
# Load local models for inference
|
11 |
+
stt_model = pipeline("automatic-speech-recognition", model="openai/whisper-small", torch_dtype=torch.float16)
|
12 |
+
conversation_model = pipeline("text-generation", model="facebook/blenderbot-400M-distill", torch_dtype=torch.float16)
|
13 |
+
tts_model = pipeline("text-to-speech", model="facebook/fastspeech2-en-ljspeech", torch_dtype=torch.float16)
|
14 |
+
|
15 |
+
# Load a pre-trained model for vector embeddings
|
16 |
+
embedding_model = SentenceTransformer('all-MiniLM-L6-v2')
|
17 |
+
|
18 |
+
# Parse PDF and create resume content
|
19 |
+
def parse_resume(pdf):
|
20 |
+
"""Extract text from an uploaded PDF file."""
|
21 |
+
reader = PyPDF2.PdfReader(pdf)
|
22 |
+
text = "\n".join(page.extract_text() for page in reader.pages if page.extract_text())
|
23 |
+
sections = {"Resume Content": text}
|
24 |
+
return sections
|
25 |
+
|
26 |
+
# Process job description text
|
27 |
+
def process_job_description(job_desc):
|
28 |
+
"""Encode the job description for analysis."""
|
29 |
+
return embedding_model.encode(job_desc)
|
30 |
+
|
31 |
+
# Process resume and generate embeddings
|
32 |
+
def process_resume(pdf):
|
33 |
+
resume_content = parse_resume(pdf)
|
34 |
+
resume_embeddings = {
|
35 |
+
section: embedding_model.encode(content) for section, content in resume_content.items()
|
36 |
+
}
|
37 |
+
return resume_embeddings
|
38 |
+
|
39 |
+
# Generate question from user response
|
40 |
+
def generate_question(user_input, resume_embeddings):
|
41 |
+
"""Find the most relevant section in the resume and generate a question."""
|
42 |
+
user_embedding = embedding_model.encode(user_input)
|
43 |
+
similarities = {
|
44 |
+
section: cosine_similarity([user_embedding], [embedding])[0][0]
|
45 |
+
for section, embedding in resume_embeddings.items()
|
46 |
+
}
|
47 |
+
most_relevant_section = max(similarities, key=similarities.get)
|
48 |
+
return f"Based on your experience in {most_relevant_section}, can you elaborate more?"
|
49 |
+
|
50 |
+
# Generate TTS output
|
51 |
+
def generate_audio(text):
|
52 |
+
"""Convert text to audio using Hugging Face TTS model."""
|
53 |
+
audio_data = tts_model(text, return_tensors=True)["waveform"]
|
54 |
+
return audio_data
|
55 |
+
|
56 |
+
# Gradio interface
|
57 |
+
class MockInterview:
|
58 |
+
def __init__(self):
|
59 |
+
self.resume_embeddings = None
|
60 |
+
self.job_desc_embedding = None
|
61 |
+
self.interview_active = False
|
62 |
+
|
63 |
+
def upload_inputs(self, resume, job_desc):
|
64 |
+
self.resume_embeddings = process_resume(resume)
|
65 |
+
self.job_desc_embedding = process_job_description(job_desc)
|
66 |
+
self.interview_active = True
|
67 |
+
question = "Tell me about yourself."
|
68 |
+
audio_output = generate_audio(question)
|
69 |
+
return "Resume and job description processed. Starting the interview.", audio_output
|
70 |
+
|
71 |
+
def conduct_interview(self, audio_file):
|
72 |
+
if not self.interview_active:
|
73 |
+
return "Please upload your resume and job description first.", None
|
74 |
+
|
75 |
+
# Transcribe audio
|
76 |
+
transcription = stt_model(audio_file)["text"]
|
77 |
+
if not transcription.strip():
|
78 |
+
return "No audio detected. Please try again.", None
|
79 |
+
|
80 |
+
# Generate next question
|
81 |
+
question = generate_question(transcription, self.resume_embeddings)
|
82 |
+
audio_output = generate_audio(question)
|
83 |
+
return transcription, audio_output
|
84 |
+
|
85 |
+
def end_interview(self):
|
86 |
+
self.interview_active = False
|
87 |
+
audio_output = generate_audio("Thank you for participating in the interview. Goodbye!")
|
88 |
+
return "Interview ended. Thank you for participating.", audio_output
|
89 |
+
|
90 |
+
mock_interview = MockInterview()
|
91 |
+
|
92 |
+
def upload_inputs(resume, job_desc):
|
93 |
+
return mock_interview.upload_inputs(resume, job_desc)
|
94 |
+
|
95 |
+
def conduct_interview(audio_file):
|
96 |
+
return mock_interview.conduct_interview(audio_file)
|
97 |
+
|
98 |
+
def end_interview():
|
99 |
+
return mock_interview.end_interview()
|
100 |
+
|
101 |
+
interface = gr.Blocks()
|
102 |
+
with interface:
|
103 |
+
gr.Markdown("""# Mock Interview AI
|
104 |
+
Upload your resume and job description, then engage in a realistic audio-based interview simulation.""")
|
105 |
+
|
106 |
+
with gr.Row():
|
107 |
+
resume_input = gr.File(label="Upload Resume (PDF)")
|
108 |
+
job_desc_input = gr.Textbox(label="Paste Job Description")
|
109 |
+
upload_button = gr.Button("Upload and Start Interview")
|
110 |
+
|
111 |
+
with gr.Row():
|
112 |
+
audio_input = gr.Audio(type="filepath", label="Respond with Your Answer")
|
113 |
+
transcription_output = gr.Textbox(label="Transcription")
|
114 |
+
question_output = gr.Audio(label="Question Audio")
|
115 |
+
submit_button = gr.Button("Submit Response")
|
116 |
+
end_button = gr.Button("End Interview")
|
117 |
+
|
118 |
+
upload_button.click(upload_inputs, inputs=[resume_input, job_desc_input], outputs=[transcription_output, question_output])
|
119 |
+
submit_button.click(conduct_interview, inputs=[audio_input], outputs=[transcription_output, question_output])
|
120 |
+
end_button.click(end_interview, outputs=[transcription_output, question_output])
|
121 |
+
|
122 |
+
if __name__ == "__main__":
|
123 |
+
interface.launch()
|