Havells × Turrant.ai
← Back to Overview

Sound-Based Fault Diagnosis

Record 10 seconds of audio - AI tells you what's wrong

The Problem

When a Havells fan starts making noise, or an AC begins humming louder than usual, the consumer's only options are: call the 1800 number (long hold times, generic troubleshooting scripts), visit a service center, or call a local electrician who may misdiagnose the issue.

  • Consumers can hear something is wrong but can't describe the fault accurately to a support agent
  • Phone support agents follow generic scripts - "have you tried turning it off and on?"
  • Unnecessary technician visits for issues that could be self-resolved (loose blade, dust buildup)
  • Misdiagnosis by local electricians leads to wrong parts ordered and repeat visits
  • No data on common failure patterns across product lines and regions
The Solution

Consumer sends a 10-second audio recording of the malfunctioning product via WhatsApp. AI analyzes the sound pattern - frequency, rhythm, amplitude - and matches it against known fault signatures. Within seconds, the consumer gets a diagnosis, severity level, and recommended next step: self-fix instruction, or technician booking with the fault pre-identified.

  • Works via WhatsApp voice note - zero friction, no app needed
  • AI trained on audio patterns of common faults: bearing wear, blade imbalance, capacitor failure, motor winding issues
  • Returns diagnosis with confidence level and recommended action
  • If self-fixable (loose blade, dust): sends video guide
  • If technician needed: books visit with fault pre-diagnosed, right parts pre-ordered
What the AI Hears

Every mechanical fault produces a distinct acoustic signature. Fans and motors are especially well-suited to audio diagnosis because their sound patterns are predictable and change in specific ways when components degrade.

Bearing Wear

Grinding or scraping sound, increases with speed

Worn ball bearings produce a distinctive high-frequency grinding. The AI detects the frequency shift relative to motor RPM to confirm bearing degradation vs. other friction sources.

Blade Imbalance

Rhythmic wobble or thumping at rotation frequency

An unbalanced blade creates a periodic vibration at the fan's rotation speed. AI detects the amplitude and frequency pattern that distinguishes imbalance from other rhythmic faults.

Capacitor Failure

Humming without rotation, or slow startup with buzzing

A failing capacitor causes the motor to hum at mains frequency (50Hz) without generating enough torque to spin. The AI recognizes this distinct hum-without-rotation pattern.

Motor Winding Issue

Electrical buzzing, overheating smell often accompanies

Inter-turn short circuits in the winding produce abnormal electrical noise at harmonic frequencies. AI flags this as urgent - continued operation risks permanent motor damage.

Ceiling Fans Table/Pedestal Fans Exhaust Fans Air Conditioners Water Purifiers (pump noise) Mixer Grinders
How It Works
  1. Consumer sends WhatsApp voice note - records 10 seconds of the product running with the fault audible
  2. Audio preprocessing - noise reduction, ambient filtering, and extraction of the product's core sound signature
  3. Spectrogram analysis - AI converts audio to frequency-time representation and matches against known fault patterns for that product category
  4. Diagnosis returned - fault type, confidence level (high/medium/low), severity (urgent/moderate/minor), and plain-language explanation
  5. Action recommendation - self-fix with video guide, or technician booking with fault and required parts pre-identified
  6. Feedback loop - technician confirms or corrects diagnosis after visit, improving the model over time

Technical Foundation

Audio ML Model: Convolutional neural network trained on spectrogram images of product sounds. Transfer learning from industrial predictive maintenance models, fine-tuned on Havells product-specific audio data.

Training Data: Initial dataset built from service center recordings - technicians record products before and after repair. Augmented with synthetic fault audio generated from known acoustic models of bearing, blade, and motor faults.

Accuracy Target: 80%+ diagnosis accuracy within 6 months of deployment, improving to 90%+ as the feedback loop from technician confirmations grows the training dataset.

Ambient Noise Handling: Indian homes have background noise (TV, traffic, other appliances). The preprocessing pipeline isolates the target product's sound using frequency band filtering specific to each product category.

Key Benefits
  • First-of-its-kind in Indian consumer electronics - no competitor offers audio-based diagnosis
  • Reduces unnecessary technician visits by 30-40% - self-fixable issues resolved instantly
  • When a technician does visit, the fault is pre-diagnosed and parts pre-ordered - first-visit resolution rate improves
  • Builds a proprietary dataset of product failure patterns - feeds into quality improvement and predictive maintenance
  • Consumer delight - "I sent a voice note and it told me exactly what's wrong" - the kind of experience people share
  • Works on WhatsApp - no app, no form, no hold time. Just record and send.