Surveillance Audio Cleanup

Your security camera, dashcam, or surveillance system recording has audio that's too noisy, muffled, or compressed to understand — and you need to improve it for review, investigation, or evidence purposes.

Surveillance systems are designed for continuous video recording over extended periods — audio is typically an afterthought. The microphones are small, omnidirectional electret capsules optimized for cost rather than quality, and the audio is recorded with heavy compression at low sample rates to minimize storage. Understanding these constraints sets realistic expectations for what cleanup can achieve. **Types of surveillance recordings and their audio characteristics:** **CCTV/IP cameras:** Indoor cameras often record at 8kHz sample rate (telephone quality, 4kHz maximum frequency) or 16kHz (HD Voice quality). Outdoor cameras have additional wind and environmental noise challenges. Video is typically stored in H.264 or H.265 containers with AAC or G.711 audio. **Dashcam recordings:** In-vehicle recordings have engine rumble (80-200Hz), road noise, wind at speed, and HVAC noise alongside conversation. Cabin conversations picked up by dashcam mics can be surprisingly intelligible if windows are closed and the car isn't traveling at high speed. **Body cameras (law enforcement, security):** Higher quality than typical CCTV — often 16-48kHz audio. The mic position (chest) creates muffling from clothing movement and handling noise. **Cleanup workflow:** 1. **Export from DVR/system:** Use your DVR's export function to save as MP4 or AVI. Some systems require proprietary player software. 2. **Assess audio content:** Open in Audacity (free) → Analyze → Plot Spectrum to see frequency distribution. If there's essentially nothing above 4kHz, you're working with 8kHz telephone-quality audio — temper expectations accordingly. 3. **AI noise removal:** Upload to Voice Isolate. AI handles stationary noise (HVAC, engine rumble) effectively. 4. **EQ boost:** Apply a presence boost at 2-3kHz (+3-4dB) and a gentle high-shelf boost at 4kHz (+2dB) to improve speech intelligibility within the available bandwidth. 5. **Normalize:** Normalize to -6dBFS (louder than broadcast standard) to make quiet speech easier to hear during review. 6. **Evidence protocol:** Always preserve the original unmodified file. Work from a copy. Document the processing steps applied. **Fundamental limits:** Speech that was below the noise floor during recording cannot be recovered. Frequency content above the recording's Nyquist frequency (half the sample rate) cannot be created. Enhancement improves intelligibility of marginally audible speech — it cannot create speech from inaudible audio.

What You Get:

• Understand CCTV audio limitations (8kHz vs 16kHz sample rates, frequency ceiling)
• AI noise removal for HVAC, engine, and ambient noise in surveillance recordings
• EQ presence boost technique for maximizing intelligibility within bandwidth constraints
• Evidence protocol: preserve originals, document processing steps