🎤 Pronunciation Heatmap

The tool uses Web Speech Recognition API (built into Chrome/Edge browsers) to analyze your pronunciation:

Analysis Process:

1. Speech-to-Text: Your recording is converted to text using AI (recognizes what words you said)
2. Word Comparison: Each word you said is compared to the target sentence words
3. Similarity Scoring: AI calculates accuracy % based on phonetic matching (pronunciation closeness)
4. Visual Heatmap: Results displayed as color-coded words (green = accurate, red = needs work)

What AI Evaluates:

• Phoneme accuracy: Individual sound pronunciation (e.g., TH sound in "think")
• Word recognition: Did AI understand which word you said?
• Clarity: How clear your speech was (mumbling = lower score)

What AI Doesn't Evaluate (yet):

• Stress patterns (which syllable you emphasize)
• Intonation (rising/falling pitch)
• Connected speech (linking words naturally)

Tip: For best accuracy, use Chrome browser, speak at medium pace, and ensure a quiet environment without background noise!

Heatmap Color Guide:

• 🟢 Green (90-100% accuracy): Excellent! Pronunciation sounds native or near-native
• 🟡 Yellow-Green (80-90%): Very good! Minor improvements possible but highly understandable
• 🟡 Yellow (70-80%): Good! Understandable with slight accent, practice for clarity
• 🟠 Orange (50-70%): Fair! Recognizable but needs work, focus practice here
• 🔴 Red (0-50%): Needs Improvement! Word was mispronounced or not recognized

Target Scores by Goal:

• Daily Communication (Travel/Shopping): 60-70% Overall Accuracy (most words orange/yellow)
• Professional Work (Job Interviews/Meetings): 75-85% Overall Accuracy (mostly yellow/green)
• IELTS Speaking Band 7+: 80-90% Overall Accuracy + 80%+ Words Correct
• Accent Reduction (Sound Native): 90%+ Overall Accuracy with 95%+ Clarity Score

Realistic Progress Timeline:

• Week 1-2: 40-60% (beginner baseline, many red words)
• Month 1: 55-70% (orange/yellow, understandable)
• Month 3: 70-80% (yellow/green, clear communication)
• Month 6+: 80-90%+ (mostly green, near-native pronunciation)

Important: Don't be discouraged by red words early on! Focus on improving 2-3 problem words per week rather than trying to fix everything at once.

The "3S" Improvement Method for Problem Words:

1. SLOW DOWN (Break into Syllables)

• Click the red/orange word, say it syllable by syllable very slowly
• Example: "pronunciation" → pro-nun-ci-A-tion (5 syllables, stress on "A")
• Repeat each syllable 5 times: "pro, pro, pro, pro, pro"
• Then gradually speed up: slow → medium → normal pace

2. SHADOW (Mimic Native Speaker)

• Play "🔊 Native Pronunciation" and pay close attention to the problem word
• Listen 3 times focusing ONLY on that word
• Shadow technique: Say the word simultaneously with the audio (overlap your voice)
• Record yourself saying just that word 5 times, then compare

3. SPOT (Identify Specific Sound Problem)

• Common problems by word:
• TH sounds ("the," "think"): Tongue touches upper teeth, practice "th-th-th-the"
• R sounds ("brown," "rough"): Tongue doesn't touch roof, lips slightly rounded
• V vs W ("very," "wet"): V uses teeth on lower lip, W uses rounded lips
• Long vs Short vowels ("ship" vs "sheep"): "Sheep" has longer "ee" sound

Daily Practice Routine for Problem Words:

1. Morning (5 min): Identify 2-3 red/orange words from yesterday's practice
2. Afternoon (5 min): Use the 3S method (Slow, Shadow, Spot) on each word
3. Evening (5 min): Re-record the full sentence and check if those words improved to yellow/green

Pro Tip: Use our Dictation Practice tool to hear problem words in different sentences at slower speeds (0.75x)!

This tool works for BOTH goals, but with different approaches:

For PRONUNCIATION CLARITY (Be Understood):

• Goal: Get 75%+ Overall Accuracy (people understand you clearly, accent is fine)
• Focus: Fix red words (major mispronunciations) first, orange words second
• Strategy: Practice Easy and Medium sentences daily
• Timeline: 1-2 months of daily practice (15-20 min/day)
• Perfect for: Job interviews, daily work communication, travel

For ACCENT REDUCTION (Sound More Native):

• Goal: Get 90%+ Overall Accuracy with 95%+ Clarity (minimal accent)
• Focus: Improve yellow/green words to perfect green, work on stress patterns
• Strategy: Practice Hard sentences, focus on connected speech and intonation
• Timeline: 3-6 months of daily practice (20-30 min/day)
• Perfect for: Acting, broadcasting, customer service, language teaching

Important Mindset Clarification:

• Accent is NOT a mistake! Indian accent, Filipino accent, etc. are valid and beautiful
• Clarity matters more than "sounding American/British" for professional success
• 75%+ accuracy means you're perfectly understandable in global business
• Only reduce accent if YOU want to, not because of pressure

Tool Limitations for Accent Reduction:

• Doesn't analyze stress patterns (which syllable is emphasized)
• Doesn't evaluate intonation (rising/falling pitch for questions/statements)
• Doesn't teach connected speech (how natives link words: "want to" → "wanna")

For complete accent reduction, combine this tool with recording practice and professional feedback on stress/intonation!

🔒 No, your voice recordings are 100% private and NOT stored!

How Privacy Works:

• Local Processing: All recording and analysis happens in your browser (client-side)
• No Upload: Your voice is NEVER sent to Skillor servers or any cloud storage
• Temporary Memory: Recording exists only in your browser's temporary memory during the session
• Automatic Deletion: When you close the browser tab, all recording data is immediately erased

Third-Party API (Google Speech Recognition):

• The Web Speech API (built into Chrome) is provided by Google
• Audio is sent to Google's servers for transcription (converting speech to text)
• According to Google's policy, this audio is NOT stored or linked to your Google account
• It's processed anonymously in real-time and immediately discarded

Data Collected:

• ✅ Stored: Your accuracy scores, sentence selection (saved in browser localStorage for progress tracking)
• ❌ NOT Stored: Voice recordings, microphone audio, personal information

For Maximum Privacy:

• Use Incognito/Private browsing mode (no data saved at all)
• Deny microphone permission after practice (browser settings)
• Practice offline (some features won't work but recording stays 100% local)

Summary: Your pronunciation practice is as private as speaking in an empty room. No one (including us) can hear your recordings!