Updateprocess_interview.py

process_interview.py

@@ -37,17 +37,17 @@ from concurrent.futures import ThreadPoolExecutor
 # Setup logging
 logging.basicConfig(level=logging.INFO)
 logger = logging.getLogger(__name__)
-logging.getLogger("nemo_logging").setLevel(logging.INFO)
-logging.getLogger("nemo").setLevel(logging.INFO)
+logging.getLogger("nemo_logging").setLevel(logging.ERROR)
+logging.getLogger("nemo").setLevel(logging.ERROR)
 
 # Configuration
-AUDIO_DIR = "./Uploads"
+AUDIO_DIR = "./uploads"
 OUTPUT_DIR = "./processed_audio"
 os.makedirs(OUTPUT_DIR, exist_ok=True)
 
 # API Keys
-PINECONE_KEY = os.getenv("PINECONE_KEY")'
-ASSEMBLYAI_KEY = 'os.getenv("ASSEMBLYAI_KEY")'
+PINECONE_KEY = os.getenv("PINECONE_KEY")
+ASSEMBLYAI_KEY = os.getenv("ASSEMBLYAI_KEY")
 GEMINI_API_KEY = os.getenv("GEMINI_API_KEY")
 
 def download_audio_from_url(url: str) -> str:
@@ -211,31 +211,31 @@ def process_utterance(utterance, full_audio, wav_file):
         else:
             speaker_id = f"unknown_{uuid.uuid4().hex[:6]}"
             speaker_name = f"Speaker_{speaker_id[-4:]}"
-            index.upsert([(speaker_id, embedding_list, {"speaker_name": speaker_id})])
+            index.upsert([(speaker_id, embedding_list, {"speaker_name": speaker_name})])
         os.remove(temp_path)
         return {
-            ...
-            **speech, 'speaker': speaker_name,
+            **utterance,
+            'speaker': speaker_name,
             'speaker_id': speaker_id,
             'embedding': embedding_list
         }
     except Exception as e:
         logger.error(f"Utterance processing failed: {str(e)}", exc_info=True)
         return {
-            ...
-            speech, 'speech': 'Unknown',
-            'speaker_id': speaker_id,
-            'embedding_id': None
+            **utterance,
+            'speaker': 'Unknown',
+            'speaker_id': 'unknown',
+            'embedding': None
         }
 
-def identify_speakers(audio: Dict, text: str) -> List[Dict]:
+def identify_speakers(transcript: Dict, wav_file: str) -> List[Dict]:
     try:
-        audio = AudioSegment.from_wav(text)
-        speakers = audio['speech']
+        full_audio = AudioSegment.from_wav(wav_file)
+        utterances = transcript['utterances']
         with ThreadPoolExecutor(max_workers=5) as executor:
             futures = [
-                executor.submit(process_speech, speech, speakers, text)
-                for speech in speakers
+                executor.submit(process_utterance, utterance, full_audio, wav_file)
+                for utterance in utterances
             ]
             results = [f.result() for f in futures]
         return results
@@ -243,33 +243,31 @@ def identify_speakers(audio: Dict, text: str) -> List[Dict]:
         logger.error(f"Speaker identification failed: {str(e)}")
         raise
 
-def train_role_classifier(speakers: List[Dict]):
+def train_role_classifier(utterances: List[Dict]):
     try:
-        speech = [u['speech'].split()]
-        vectorizer = TfidfVectorizer(max_features=500, ngram_range=(1,2))
-        X_text = vectorizer.fit_transform(speech)
+        texts = [u['text'] for u in utterances]
+        vectorizer = TfidfVectorizer(max_features=500, ngram_range=(1, 2))
+        X_text = vectorizer.fit_transform(texts)
         features = []
         labels = []
-        for i, speaker in enumerate(speakers):
-            utterance = speaker['speech_features']
+        for i, utterance in enumerate(utterances):
+            prosodic = utterance['prosodic_features']
             feat = [
-                utterance['duration'], utterance['speech_rate'], utterance['duration'], utterance['mean_pitch'], 
-                utterance['min_pitch'], utterance['max_pitch'], 
-                utterance['speech_sd'], utterance['intensityLevel'], 
-                utterance['intensity_level'], 
-                utterance['speechMax']], utterance['speechSD'],
+                prosodic['duration'], prosodic['mean_pitch'], prosodic['min_pitch'],
+                prosodic['max_pitch'], prosodic['pitch_sd'], prosodic['intensityMean'],
+                prosodic['intensityMin'], prosodic['intensityMax'], prosodic['intensitySD'],
             ]
-            feat.extend(X_text[i].toarray()[0])
-            doc = nlp(speaker['speech'])
-            speech.extend([
-                int(speaker['speech'].endswith('?'))),
-                len(re.findall(r'\b(why|how|what|when|where|who|which)\b', speaker['speech'].lower())),
-                len(speaker['speech'].split())),
-                sum(frequency for token in speech if token.pos_ == 'VERB'),
-                sum(frequency for token in speech if token.pos == 'NOUN')
+            feat.extend(X_text[i].toarray()[0].tolist())
+            doc = nlp(utterance['text'])
+            feat.extend([
+                int(utterance['text'].endswith('?')),
+                len(re.findall(r'\b(why|how|what|when|where|who|which)\b', utterance['text'].lower())),
+                len(utterance['text'].split()),
+                sum(1 for token in doc if token.pos_ == 'VERB'),
+                sum(1 for token in doc if token.pos_ == 'NOUN')
             ])
             features.append(feat)
-            labels.append((0 if i % 2 == 0 else 1))
+            labels.append(0 if i % 2 == 0 else 1)
         scaler = StandardScaler()
         X = scaler.fit_transform(features)
         clf = RandomForestClassifier(
@@ -284,53 +282,53 @@ def train_role_classifier(speakers: List[Dict]):
         logger.error(f"Classifier training failed: {str(e)}")
         raise
 
-def classify_roles(speakers: List[Dict], clf, vectorizer, scaler):
+def classify_roles(utterances: List[Dict], clf, vectorizer, scaler):
     try:
-        speech = [u['speech'] for u in speakers]
-        X_text = vectorizer.transform(speech)
+        texts = [u['text'] for u in utterances]
+        X_text = vectorizer.transform(texts)
         results = []
-        for i, speaker in enumerate(speakers):
-            prosodic = speaker['speech_features']
+        for i, utterance in enumerate(utterances):
+            prosodic = utterance['prosodic_features']
             feat = [
                 prosodic['duration'], prosodic['mean_pitch'], prosodic['min_pitch'],
                 prosodic['max_pitch'], prosodic['pitch_sd'], prosodic['intensityMean'],
                 prosodic['intensityMin'], prosodic['intensityMax'], prosodic['intensitySD'],
             ]
             feat.extend(X_text[i].toarray()[0].tolist())
-            doc = nlp(speaker['speech'])
+            doc = nlp(utterance['text'])
             feat.extend([
-                int(speaker['speech'].endswith('?')),
-                len(re.findall(r'\b(why|how|what|when|where|who|which)\b', speaker['speech'].lower())),
-                len(speaker['speech'].split()),
+                int(utterance['text'].endswith('?')),
+                len(re.findall(r'\b(why|how|what|when|where|who|which)\b', utterance['text'].lower())),
+                len(utterance['text'].split()),
                 sum(1 for token in doc if token.pos_ == 'VERB'),
                 sum(1 for token in doc if token.pos_ == 'NOUN')
             ])
             X = scaler.transform([feat])
             role = 'Interviewer' if clf.predict(X)[0] == 0 else 'Interviewee'
-            results.append({**speaker, 'role': role})
+            results.append({**utterance, 'role': role})
         return results
     except Exception as e:
         logger.error(f"Role classification failed: {str(e)}")
         raise
 
-def analyze_interviewee_voice(audio_path: str, speakers: List[Dict]) -> Dict:
+def analyze_interviewee_voice(audio_path: str, utterances: List[Dict]) -> Dict:
     try:
         y, sr = librosa.load(audio_path, sr=16000)
-        interviewee_speakers = [u for u in speakers if u['role'] == 'Interviewee']
-        if not interviewee_speakers:
-            return {'error': 'No interviewee speeches found'}
+        interviewee_utterances = [u for u in utterances if u['role'] == 'Interviewee']
+        if not interviewee_utterances:
+            return {'error': 'No interviewee utterances found'}
         segments = []
-        for u in interviewee_speakers:
+        for u in interviewee_utterances:
             start = int(u['start'] * sr / 1000)
             end = int(u['end'] * sr / 1000)
             segments.append(y[start:end])
-        total_duration = sum(u['speech_features']['duration'] for u in interviewee_speakers)
-        total_words = sum(len(u['speech'].split()) for u in interviewee_speakers)
+        total_duration = sum(u['prosodic_features']['duration'] for u in interviewee_utterances)
+        total_words = sum(len(u['text'].split()) for u in interviewee_utterances)
         speaking_rate = total_words / total_duration if total_duration > 0 else 0
         filler_words = ['um', 'uh', 'like', 'you know', 'so', 'i mean']
-        filler_count = sum(sum(u['speech'].lower().count(fw) for fw in filler_words) for u in interviewee_speakers)
+        filler_count = sum(sum(u['text'].lower().count(fw) for fw in filler_words) for u in interviewee_utterances)
         filler_ratio = filler_count / total_words if total_words > 0 else 0
-        all_words = ' '.join(u['speech'].lower() for u in interviewee_speakers).split()
+        all_words = ' '.join(u['text'].lower() for u in interviewee_utterances).split()
         word_counts = {}
         for i in range(len(all_words) - 1):
             bigram = (all_words[i], all_words[i + 1])
@@ -374,19 +372,19 @@ def generate_voice_interpretation(analysis: Dict) -> str:
         return "Voice analysis unavailable due to processing limitations."
     interpretation_lines = [
         "Vocal Performance Profile:",
-        f"- Speaking Rate: {analysis['speaking_rate']} words/sec - Benchmark: 2.0-3.0 wps for clear delivery",
-        f"- Filler Word Frequency: {analysis['filler_ratio'] * 100:.1f}% - Measures non-content words",
-        f"- Repetition Index: {analysis['repetition_score']:.3f} - Frequency of repeated phrases",
-        f"- Anxiety Indicator: {analysis['interpretation']['anxiety_level']} (Score: {analysis['composite_scores']['anxiety']:.3f}) - Pitch and vocal stability",
-        f"- Confidence Indicator: {analysis['interpretation']['confidence_level']} (Score: {analysis['composite_scores']['confidence']:.3f}) - Vocal strength",
-        f"- Fluency Rating: {analysis['interpretation']['fluency_level']} - Speech flow and coherence",
+        f"- Speaking Rate: {analysis['speaking_rate']} words/sec - Benchmark: 2.0-3.0 wps for clear, professional delivery",
+        f"- Filler Word Frequency: {analysis['filler_ratio'] * 100:.1f}% - Measures non-content words (e.g., 'um', 'like')",
+        f"- Repetition Index: {analysis['repetition_score']:.3f} - Frequency of repeated phrases or ideas",
+        f"- Anxiety Indicator: {analysis['interpretation']['anxiety_level']} (Score: {analysis['composite_scores']['anxiety']:.3f}) - Derived from pitch variation and vocal stability",
+        f"- Confidence Indicator: {analysis['interpretation']['confidence_level']} (Score: {analysis['composite_scores']['confidence']:.3f}) - Reflects vocal strength and consistency",
+        f"- Fluency Rating: {analysis['interpretation']['fluency_level']} - Assesses speech flow and coherence",
         "",
-        "HR Insights:",
-        "- Rapid speech (>3.0 wps) may signal enthusiasm but risks clarity.",
-        "- High filler word use reduces perceived professionalism.",
-        "- Elevated anxiety suggests pressure; training can build resilience.",
-        "- Strong confidence aligns with leadership presence.",
-        "- Fluent speech enhances engagement, critical for team roles."
+        "HR Performance Insights:",
+        "- Rapid speech (>3.0 wps) may signal enthusiasm but risks clarity; slower, deliberate pacing enhances professionalism.",
+        "- Elevated filler word use reduces perceived polish and can distract from key messages.",
+        "- High anxiety scores suggest interview pressure; training can build resilience.",
+        "- Strong confidence indicators align with leadership presence and effective communication.",
+        "- Fluent speech enhances engagement, critical for client-facing or team roles."
     ]
     return "\n".join(interpretation_lines)
 
@@ -394,18 +392,18 @@ def generate_anxiety_confidence_chart(composite_scores: Dict, chart_path_or_buff
     try:
         labels = ['Anxiety', 'Confidence']
         scores = [composite_scores.get('anxiety', 0), composite_scores.get('confidence', 0)]
-        fig, ax = plt.subplots(figsize=(5, 3.5))
-        bars = ax.bar(labels, scores, color=['#FF5252', '#26A69A'], edgecolor='black', width=0.45)
+        fig, ax = plt.subplots(figsize=(5, 3))
+        bars = ax.bar(labels, scores, color=['#FF6B6B', '#4ECDC4'], edgecolor='black', width=0.6)
         ax.set_ylabel('Score (Normalized)', fontsize=12)
         ax.set_title('Vocal Dynamics: Anxiety vs. Confidence', fontsize=14, pad=15)
-        ax.set_ylim(0, 1.3)
+        ax.set_ylim(0, 1.2)
         for bar in bars:
             height = bar.get_height()
             ax.text(bar.get_x() + bar.get_width()/2, height + 0.05, f"{height:.2f}",
                     ha='center', color='black', fontweight='bold', fontsize=11)
         ax.grid(True, axis='y', linestyle='--', alpha=0.7)
         plt.tight_layout()
-        plt.savefig(chart_path_or_buffer, format='png', bbox_inches='tight', dpi=300)
+        plt.savefig(chart_path_or_buffer, format='png', bbox_inches='tight', dpi=200)
         plt.close(fig)
     except Exception as e:
         logger.error(f"Error generating chart: {str(e)}")
@@ -449,29 +447,29 @@ def generate_report(analysis_data: Dict) -> str:
             elif acceptance_prob >= 40: acceptance_line += "HR Verdict: Moderate potential, requires additional assessment and skill-building."
             else: acceptance_line += "HR Verdict: Limited fit, significant improvement needed for role alignment."
         prompt = f"""
-        You are EvalBot, a senior HR consultant with 20+ years of experience, delivering a polished, concise, and engaging interview analysis report. Use a professional tone, clear headings, and bullet points ('- ') for readability. Avoid redundancy and ensure distinct sections for strengths, growth areas, and recommendations.
+        You are EvalBot, a senior HR consultant with 20+ years of experience, delivering a polished, concise, and visually engaging interview analysis report. Use a professional tone, clear headings, and bullet points ('- ') for readability. Focus on candidate suitability, strengths, and actionable growth strategies.
         {acceptance_line}
         **1. Executive Summary**
-        - Provide a concise overview of performance, key metrics, and hiring potential.
+        - Deliver a crisp overview of the candidate's performance, emphasizing key metrics and hiring potential.
         - Interview length: {analysis_data['text_analysis']['total_duration']:.2f} seconds
         - Speaker turns: {analysis_data['text_analysis']['speaker_turns']}
         - Participants: {', '.join(analysis_data['speakers'])}
         **2. Communication and Vocal Dynamics**
-        - Evaluate vocal delivery (rate, fluency, confidence) and professional impact.
-        - Offer HR insights on workplace alignment.
+        - Assess the candidate's vocal delivery (rate, fluency, confidence) and its impact on professional presence.
+        - Provide HR insights on how these traits align with workplace expectations.
         {voice_interpretation}
         **3. Competency and Content Evaluation**
-        - Assess competencies: leadership, problem-solving, communication, adaptability.
-        - List strengths and growth areas separately, with specific examples.
+        - Evaluate responses for core competencies: leadership, problem-solving, communication, adaptability.
+        - Highlight strengths and growth areas with specific, concise examples.
         - Sample responses:
         {chr(10).join(interviewee_responses)}
         **4. Role Fit and Growth Potential**
-        - Analyze cultural fit, role readiness, and long-term potential.
-        - Highlight enthusiasm and scalability.
+        - Analyze alignment with professional roles, focusing on cultural fit, readiness, and scalability.
+        - Consider enthusiasm, teamwork, and long-term potential.
         **5. Strategic HR Recommendations**
-        - Provide distinct, prioritized strategies for candidate growth.
-        - Target: Communication, Response Depth, Professional Presence.
-        - List clear next steps for hiring managers (e.g., advance, train, assess).
+        - Offer prioritized, actionable strategies to enhance candidate performance.
+        - Target: Communication Effectiveness, Response Depth, Professional Impact.
+        - Suggest clear next steps for hiring managers (e.g., advance, train, assess).
         """
         response = gemini_model.generate_content(prompt)
         return response.text
@@ -482,40 +480,41 @@ def generate_report(analysis_data: Dict) -> str:
 def create_pdf_report(analysis_data: Dict, output_path: str, gemini_report_text: str):
     try:
         doc = SimpleDocTemplate(output_path, pagesize=letter,
-                                rightMargin=0.7*inch, leftMargin=0.7*inch,
-                                topMargin=0.9*inch, bottomMargin=0.9*inch)
+                                rightMargin=0.6*inch, leftMargin=0.6*inch,
+                                topMargin=0.8*inch, bottomMargin=0.8*inch)
         styles = getSampleStyleSheet()
-        h1 = ParagraphStyle(name='Heading1', fontSize=22, leading=26, spaceAfter=20, alignment=1, textColor=colors.HexColor('#003087'), fontName='Helvetica-Bold')
-        h2 = ParagraphStyle(name='Heading2', fontSize=15, leading=18, spaceBefore=14, spaceAfter=8, textColor=colors.HexColor('#0050BC'), fontName='Helvetica-Bold')
-        h3 = ParagraphStyle(name='Heading3', fontSize=11, leading=14, spaceBefore=10, spaceAfter=6, textColor=colors.HexColor('#3F7CFF'), fontName='Helvetica')
-        body_text = ParagraphStyle(name='BodyText', fontSize=10, leading=13, spaceAfter=8, fontName='Helvetica', textColor=colors.HexColor('#333333'))
-        bullet_style = ParagraphStyle(name='Bullet', parent=body_text, leftIndent=20, bulletIndent=10, fontName='Helvetica', bulletFontName='Helvetica', bulletFontSize=10)
+        h1 = ParagraphStyle(name='Heading1', fontSize=24, leading=28, spaceAfter=25, alignment=1, textColor=colors.HexColor('#1A3C5E'), fontName='Helvetica-Bold')
+        h2 = ParagraphStyle(name='Heading2', fontSize=16, leading=20, spaceBefore=16, spaceAfter=10, textColor=colors.HexColor('#2E5A87'), fontName='Helvetica-Bold')
+        h3 = ParagraphStyle(name='Heading3', fontSize=12, leading=16, spaceBefore=12, spaceAfter=8, textColor=colors.HexColor('#4A6FA5'), fontName='Helvetica')
+        body_text = ParagraphStyle(name='BodyText', parent=styles['Normal'], fontSize=10, leading=14, spaceAfter=10, fontName='Helvetica')
+        bullet_style = ParagraphStyle(name='Bullet', parent=body_text, leftIndent=25, bulletIndent=12, fontName='Helvetica')
         
         story = []
 
         def header_footer(canvas, doc):
             canvas.saveState()
-            canvas.setFont('Helvetica', 8)
+            canvas.setFont('Helvetica', 9)
+            canvas.setFillColor(colors.HexColor('#666666'))
+            canvas.drawString(doc.leftMargin, 0.5 * inch, f"Page {doc.page} | EvalBot HR Interview Report | Confidential")
+            canvas.setStrokeColor(colors.HexColor('#2E5A87'))
+            canvas.setLineWidth(1.2)
+            canvas.line(doc.leftMargin, doc.height + 0.9*inch, doc.width + doc.leftMargin, doc.height + 0.9*inch)
+            canvas.setFont('Helvetica-Bold', 11)
+            canvas.drawString(doc.leftMargin, doc.height + 0.95*inch, "Candidate Interview Analysis")
             canvas.setFillColor(colors.HexColor('#666666'))
-            canvas.drawString(doc.leftMargin, 0.4 * inch, f"Page {doc.page} | EvalBot HR Interview Report | Confidential")
-            canvas.setStrokeColor(colors.HexColor('#0050BC'))
-            canvas.setLineWidth(1)
-            canvas.line(doc.leftMargin, doc.height + 0.85*inch, doc.width + doc.leftMargin, doc.height + 0.85*inch)
-            canvas.setFont('Helvetica-Bold', 10)
-            canvas.drawString(doc.leftMargin, doc.height + 0.9*inch, "Candidate Interview Analysis")
-            canvas.drawRightString(doc.width + doc.leftMargin, doc.height + 0.9*inch, time.strftime('%B %d, %Y'))
+            canvas.drawRightString(doc.width + doc.leftMargin, doc.height + 0.95*inch, time.strftime('%B %d, %Y'))
             canvas.restoreState()
 
         # Title Page
         story.append(Paragraph("Candidate Interview Analysis", h1))
-        story.append(Paragraph(f"Generated: {time.strftime('%B %d, %Y')}", ParagraphStyle(name='Date', alignment=1, fontSize=10, textColor=colors.HexColor('#666666'), fontName='Helvetica')))
-        story.append(Spacer(1, 0.5 * inch))
+        story.append(Paragraph(f"Generated: {time.strftime('%B %d, %Y')}", ParagraphStyle(name='Date', alignment=1, fontSize=11, textColor=colors.HexColor('#666666'), fontName='Helvetica')))
+        story.append(Spacer(1, 0.6 * inch))
         acceptance_prob = analysis_data.get('acceptance_probability')
         if acceptance_prob is not None:
-            story.append(Paragraph("Hiring Suitability Snapshot", h2))
+            story.append(Paragraph("Hiring Suitability Overview", h2))
             prob_color = colors.HexColor('#2E7D32') if acceptance_prob >= 80 else (colors.HexColor('#F57C00') if acceptance_prob >= 60 else colors.HexColor('#D32F2F'))
-            story.append(Paragraph(f"Suitability Score: <font size=16 color='{prob_color.hexval()}'><b>{acceptance_prob:.2f}%</b></font>",
-                                 ParagraphStyle(name='Prob', fontSize=12, spaceAfter=12, alignment=1, fontName='Helvetica-Bold')))
+            story.append(Paragraph(f"Hiring Suitability Score: <font size=18 color='{prob_color.hexval()}'><b>{acceptance_prob:.2f}%</b></font>",
+                                 ParagraphStyle(name='Prob', fontSize=14, spaceAfter=15, alignment=1, fontName='Helvetica-Bold')))
             if acceptance_prob >= 80:
                 story.append(Paragraph("<b>HR Verdict:</b> Outstanding candidate, highly recommended for immediate advancement.", body_text))
             elif acceptance_prob >= 60:
@@ -524,102 +523,89 @@ def create_pdf_report(analysis_data: Dict, output_path: str, gemini_report_text:
                 story.append(Paragraph("<b>HR Verdict:</b> Moderate potential, requires additional assessment and skill-building.", body_text))
             else:
                 story.append(Paragraph("<b>HR Verdict:</b> Limited fit, significant improvement needed for role alignment.", body_text))
-            story.append(Spacer(1, 0.3 * inch))
+            story.append(Spacer(1, 0.4 * inch))
             table_data = [
-                ['Metric', 'Value'],
-                ['Interview Duration', f"{analysis_data['text_analysis']['total_duration']:.2f} seconds"],
+                ['Key Metrics', 'Value'],
+                ['Interview Length', f"{analysis_data['text_analysis']['total_duration']:.2f} seconds"],
                 ['Speaker Turns', f"{analysis_data['text_analysis']['speaker_turns']}"],
-                ['Participants', ', '.join(sorted(analysis_data['speakers']))]
+                ['Participants', ', '.join(analysis_data['speakers'])]
             ]
-            table = Table(table_data, colWidths=[2.2*inch, 3.8*inch])
+            table = Table(table_data, colWidths=[2.5*inch, 4*inch])
             table.setStyle(TableStyle([
-                ('BACKGROUND', (0,0), (-1,0), colors.HexColor('#0050BC')),
-                ('TEXTCOLOR', (0,0), (-1,0), colors.white),
+                ('BACKGROUND', (0,0), (-1,0), colors.HexColor('#2E5A87')),
+                ('TEXTCOLOR', (0,0), (-1,0), colors.whitesmoke),
                 ('ALIGN', (0,0), (-1,-1), 'LEFT'),
                 ('VALIGN', (0,0), (-1,-1), 'MIDDLE'),
-                ('FONTNAME', (0,0), (-1,0), 'Helvetica-Bold'),
-                ('FONTSIZE', (0,0), (-1,-1), 9),
-                ('BOTTOMPADDING', (0,0), (-1,0), 10),
-                ('TOPPADDING', (0,0), (-1,0), 10),
-                ('BACKGROUND', (0,1), (-1,-1), colors.HexColor('#F5F6FA')),
-                ('GRID', (0,0), (-1,-1), 0.5, colors.HexColor('#DDE4EB'))
+                ('FONTNAME', (0, 0), (-1, 0), 'Helvetica-Bold'),
+                ('FONTSIZE', (0, 0), (-1, -1), 10),
+                ('BOTTOMPADDING', (0, 0), (-1, 0), 12),
+                ('TOPPADDING', (0, 0), (-1, 0), 12),
+                ('BACKGROUND', (0, 1), (-1, -1), colors.HexColor('#F5F7FA')),
+                ('GRID', (0,0), (-1,-1), 1, colors.HexColor('#DDE4EB'))
             ]))
             story.append(table)
-        story.append(Spacer(1, 0.4 * inch))
-        story.append(Paragraph("Prepared by: EvalBot - AI-Powered HR Analysis", body_text))
+        story.append(Spacer(1, 0.5 * inch))
+        story.append(Paragraph("Prepared by: EvalBot - AI-Powered HR Analysis System", body_text))
         story.append(PageBreak())
 
         # Detailed Analysis
-        story.append(Paragraph("Detailed Candidate Evaluation", h1))
+        story.append(Paragraph("Detailed Candidate Profile", h1))
         
-        # Communication and Vocal Dynamics
         story.append(Paragraph("1. Communication & Vocal Dynamics", h2))
         voice_analysis = analysis_data.get('voice_analysis', {})
         if voice_analysis and 'error' not in voice_analysis:
             table_data = [
                 ['Metric', 'Value', 'HR Insight'],
-                ['Speaking Rate', f"{voice_analysis.get('speaking_rate', 0):.2f} words/sec", 'Benchmark: 2.0-3.0 wps; impacts clarity'],
-                ['Filler Words', f"{voice_analysis.get('filler_ratio', 0) * 100:.1f}%", 'High usage reduces credibility'],
-                ['Anxiety', voice_analysis.get('interpretation', {}).get('anxiety_level', 'N/A'), f"Score: {voice_analysis.get('composite_scores', {}).get('anxiety', 0):.3f}; stress response"],
-                ['Confidence', voice_analysis.get('interpretation', {}).get('confidence_level', 'N/A'), f"Score: {voice_analysis.get('composite_scores', {}).get('confidence', 0):.3f}; vocal strength"],
-                ['Fluency', voice_analysis.get('interpretation', {}).get('fluency_level', 'N/A'), 'Drives engagement']
+                ['Speaking Rate', f"{voice_analysis.get('speaking_rate', 0):.2f} words/sec", 'Benchmark: 2.0-3.0 wps; affects clarity, poise'],
+                ['Filler Word Frequency', f"{voice_analysis.get('filler_ratio', 0) * 100:.1f}%", 'Excess use impacts polish, credibility'],
+                ['Anxiety Indicator', voice_analysis.get('interpretation', {}).get('anxiety_level', 'N/A'), f"Score: {voice_analysis.get('composite_scores', {}).get('anxiety', 0):.3f}; shows stress response"],
+                ['Confidence Indicator', voice_analysis.get('interpretation', {}).get('confidence_level', 'N/A'), f"Score: {voice_analysis.get('composite_scores', {}).get('confidence', 0):.3f}; reflects vocal strength"],
+                ['Fluency Rating', voice_analysis.get('interpretation', {}).get('fluency_level', 'N/A'), 'Drives engagement, message impact']
             ]
-            table = Table(table_data, colWidths=[1.7*inch, 1.2*inch, 3.1*inch])
+            table = Table(table_data, colWidths=[1.9*inch, 1.3*inch, 3.3*inch])
             table.setStyle(TableStyle([
-                ('BACKGROUND', (0,0), (-1,0), colors.HexColor('#0050BC')),
-                ('TEXTCOLOR', (0,0), (-1,0), colors.white),
+                ('BACKGROUND', (0,0), (-1,0), colors.HexColor('#2E5A87')),
+                ('TEXTCOLOR', (0,0), (-1,0), colors.whitesmoke),
                 ('ALIGN', (0,0), (-1,-1), 'LEFT'),
                 ('VALIGN', (0,0), (-1,-1), 'MIDDLE'),
-                ('FONTNAME', (0,0), (-1,0), 'Helvetica-Bold'),
-                ('FONTSIZE', (0,0), (-1,-1), 9),
-                ('BOTTOMPADDING', (0,0), (-1,0), 10),
-                ('TOPPADDING', (0,0), (-1,0), 10),
-                ('BACKGROUND', (0,1), (-1,-1), colors.HexColor('#F5F6FA')),
-                ('GRID', (0,0), (-1,-1), 0.5, colors.HexColor('#DDE4EB'))
+                ('FONTNAME', (0, 0), (-1, 0), 'Helvetica-Bold'),
+                ('FONTSIZE', (0, 0), (-1, -1), 9),
+                ('BOTTOMPADDING', (0, 0), (-1, 0), 12),
+                ('TOPPADDING', (0, 0), (-1, 0), 12),
+                ('BACKGROUND', (0, 1), (-1, -1), colors.HexColor('#F5F7FA')),
+                ('GRID', (0,0), (-1,-1), 1, colors.HexColor('#DDE4EB'))
             ]))
             story.append(table)
-            story.append(Spacer(1, 0.2 * inch))
+            story.append(Spacer(1, 0.3 * inch))
             chart_buffer = io.BytesIO()
             generate_anxiety_confidence_chart(voice_analysis.get('composite_scores', {}), chart_buffer)
             chart_buffer.seek(0)
-            img = Image(chart_buffer, width=4.8*inch, height=3.2*inch)
+            img = Image(chart_buffer, width=5*inch, height=3*inch)
             img.hAlign = 'CENTER'
             story.append(img)
         else:
-            story.append(Paragraph("Vocal analysis unavailable.", body_text))
-        story.append(Spacer(1, 0.3 * inch))
+            story.append(Paragraph("Vocal analysis unavailable due to processing constraints.", body_text))
+        story.append(Spacer(1, 0.4 * inch))
 
         # Parse Gemini Report
-        sections = {
-            "Executive Summary": [],
-            "Communication and Vocal Dynamics": [],
-            "Competency and Content Evaluation": {"Strengths": [], "Growth Areas": []},
-            "Role Fit and Growth Potential": [],
-            "Strategic HR Recommendations": {"Development Priorities": [], "Next Steps": []}
-        }
+        sections = {}
+        section_titles = ["Executive Summary", "Communication and Vocal Dynamics", 
+                         "Competency and Content Evaluation", 
+                         "Role Fit and Growth Potential", "Strategic HR Recommendations"]
+        for title in section_titles:
+            sections[title] = []
         report_parts = re.split(r'(\s*\*\*\s*\d\.\s*.*?\s*\*\*)', gemini_report_text)
         current_section = None
         for part in report_parts:
             if not part.strip(): continue
             is_heading = False
-            for title in sections.keys():
+            for title in section_titles:
                 if title.lower() in part.lower():
                     current_section = title
                     is_heading = True
                     break
             if not is_heading and current_section:
-                if current_section == "Competency and Content Evaluation":
-                    if 'strength' in part.lower() or any(k in part.lower() for k in ['leadership', 'problem-solving', 'communication', 'adaptability']):
-                        sections[current_section]["Strengths"].append(part.strip())
-                    elif 'improve' in part.lower() or 'grow' in part.lower() or 'challenge' in part.lower():
-                        sections[current_section]["Growth Areas"].append(part.strip())
-                elif current_section == "Strategic HR Recommendations":
-                    if any(k in part.lower() for k in ['communication', 'depth', 'presence', 'improve']):
-                        sections[current_section]["Development Priorities"].append(part.strip())
-                    elif any(k in part.lower() for k in ['advance', 'train', 'assess', 'next step']):
-                        sections[current_section]["Next Steps"].append(part.strip())
-                else:
-                    sections[current_section].append(part.strip())
+                sections[current_section].append(part.strip())
 
         # Executive Summary
         story.append(Paragraph("2. Executive Summary", h2))
@@ -630,28 +616,35 @@ def create_pdf_report(analysis_data: Dict, output_path: str, gemini_report_text:
                 else:
                     story.append(Paragraph(line, body_text))
         else:
-            story.append(Paragraph("Summary unavailable.", body_text))
-        story.append(Spacer(1, 0.3 * inch))
+            story.append(Paragraph("Executive summary unavailable.", body_text))
+        story.append(Spacer(1, 0.4 * inch))
 
         # Competency and Content
-        story.append(Paragraph("3. Competency & Content", h2))
-        story.append(Paragraph("Strengths", h3))
-        if sections['Competency and Content Evaluation']['Strengths']:
-            for line in sections['Competency and Content Evaluation']['Strengths']:
-                story.append(Paragraph(line.lstrip('-•* ').strip(), bullet_style))
-        else:
-            story.append(Paragraph("No strengths identified.", body_text))
-        story.append(Spacer(1, 0.2 * inch))
-        story.append(Paragraph("Growth Areas", h3))
-        if sections['Competency and Content Evaluation']['Growth Areas']:
-            for line in sections['Competency and Content Evaluation']['Growth Areas']:
-                story.append(Paragraph(line.lstrip('-•* ').strip(), bullet_style))
+        story.append(Paragraph("3. Competency & Content Evaluation", h2))
+        if sections['Competency and Content Evaluation']:
+            story.append(Paragraph("Strengths", h3))
+            strengths_found = False
+            for line in sections['Competency and Content Evaluation']:
+                if 'strength' in line.lower() or any(k in line.lower() for k in ['leadership', 'problem-solving', 'communication', 'adaptability']):
+                    story.append(Paragraph(line.lstrip('-•* ').strip(), bullet_style))
+                    strengths_found = True
+            if not strengths_found:
+                story.append(Paragraph("No specific strengths identified.", body_text))
+            story.append(Spacer(1, 0.2 * inch))
+            story.append(Paragraph("Growth Areas", h3))
+            growth_found = False
+            for line in sections['Competency and Content Evaluation']:
+                if 'improve' in line.lower() or 'weak' in line.lower() or 'challenge' in line.lower():
+                    story.append(Paragraph(line.lstrip('-•* ').strip(), bullet_style))
+                    growth_found = True
+            if not growth_found:
+                story.append(Paragraph("No specific growth areas identified.", body_text))
         else:
-            story.append(Paragraph("No growth areas identified.", body_text))
-        story.append(Spacer(1, 0.3 * inch))
+            story.append(Paragraph("Competency and content evaluation unavailable.", body_text))
+        story.append(PageBreak())
 
         # Role Fit
-        story.append(Paragraph("4. Role Fit & Potential", h2))
+        story.append(Paragraph("4. Role Fit & Growth Potential", h2))
         if sections['Role Fit and Growth Potential']:
             for line in sections['Role Fit and Growth Potential']:
                 if line.startswith(('-', '•', '*')):
@@ -659,31 +652,38 @@ def create_pdf_report(analysis_data: Dict, output_path: str, gemini_report_text:
                 else:
                     story.append(Paragraph(line, body_text))
         else:
-            story.append(Paragraph("Fit and potential analysis unavailable.", body_text))
-        story.append(Spacer(1, 0.3 * inch))
+            story.append(Paragraph("Role fit and potential analysis unavailable.", body_text))
+        story.append(Spacer(1, 0.4 * inch))
 
-        # Strategic Recommendations
-        story.append(Paragraph("5. Strategic Recommendations", h2))
-        story.append(Paragraph("Development Priorities", h3))
-        if sections['Strategic HR Recommendations']['Development Priorities']:
-            for line in sections['Strategic HR Recommendations']['Development Priorities']:
-                story.append(Paragraph(line.lstrip('-•* ').strip(), bullet_style))
-        else:
-            story.append(Paragraph("No development priorities specified.", body_text))
-        story.append(Spacer(1, 0.2 * inch))
-        story.append(Paragraph("Next Steps for Managers", h3))
-        if sections['Strategic HR Recommendations']['Next Steps']:
-            for line in sections['Strategic HR Recommendations']['Next Steps']:
-                story.append(Paragraph(line.lstrip('-•* ').strip(), bullet_style))
+        # HR Recommendations
+        story.append(Paragraph("5. Strategic HR Recommendations", h2))
+        if sections['Strategic HR Recommendations']:
+            story.append(Paragraph("Development Priorities", h3))
+            dev_found = False
+            for line in sections['Strategic HR Recommendations']:
+                if any(k in line.lower() for k in ['communication', 'clarity', 'depth', 'presence', 'improve']):
+                    story.append(Paragraph(line.lstrip('-•* ').strip(), bullet_style))
+                    dev_found = True
+            if not dev_found:
+                story.append(Paragraph("No development priorities specified.", body_text))
+            story.append(Spacer(1, 0.2 * inch))
+            story.append(Paragraph("Next Steps for Hiring Managers", h3))
+            steps_found = False
+            for line in sections['Strategic HR Recommendations']:
+                if any(k in line.lower() for k in ['advance', 'train', 'assess', 'next step']):
+                    story.append(Paragraph(line.lstrip('-•* ').strip(), bullet_style))
+                    steps_found = True
+            if not steps_found:
+                story.append(Paragraph("No specific next steps provided.", body_text))
         else:
-            story.append(Paragraph("No next steps provided.", body_text))
+            story.append(Paragraph("Strategic recommendations unavailable.", body_text))
         story.append(Spacer(1, 0.3 * inch))
-        story.append(Paragraph("This report provides a data-driven evaluation to guide hiring and development decisions.", body_text))
+        story.append(Paragraph("This report delivers a comprehensive, data-driven evaluation to guide hiring decisions and candidate development.", body_text))
 
         doc.build(story, onFirstPage=header_footer, onLaterPages=header_footer)
         return True
     except Exception as e:
-        logger.error(f"PDF creation failed: {str(e)}", exc_info=True)
+        logger.error(f"Enhanced PDF creation failed: {str(e)}", exc_info=True)
         return False
 
 def convert_to_serializable(obj):