How to Clean Up a Lecture or Conference Recording
Lecture recordings serve critical educational purposes — students review them for exams, online learners may only have the recording, people with learning differences need clear audio to process information effectively, and institutions archive them for future reference.
But lecture recordings are notoriously poor quality. Classroom and conference hall acoustics were designed for live listening, not recording. Microphones are often positioned poorly. Audience noise bleeds in. The result is audio that's technically present but frustrating to listen to.
This guide covers how to clean up lecture and conference recordings for clear, accessible audio.
Why Lecture Recordings Sound the Way They Do
Room acoustics: Lecture halls and classrooms have high ceilings, hard floors, and large glass windows — all highly reflective. The reverb time in a typical lecture hall ranges from 0.5 to 1.5 seconds. This echo is essential for live listening (it helps fill the room and makes the lecturer audible throughout) but becomes a problem in recordings.
Microphone placement: Built-in room microphones are often ceiling-mounted for wide coverage. The trade-off: they capture the entire room including background noise and use maximum gain, which elevates noise.
HVAC: Lecture rooms need to circulate air for occupant comfort — air handling systems run throughout the lecture. Their broadband noise is captured clearly by any microphone.
Audience noise: Typing, writing, shuffling, coughing, whispering — an audience of 20–200 people generates significant background noise.
Recording method: Many lecture recordings are made with a simple room microphone, smartphone on a desk, or automated capture systems with minimal optimization.
Before You Process: Organize the Recording
Cut out dead time: Most lectures include time before the lecturer starts and after they finish. Trim these sections.
Mark important sections: If you have a long lecture (60–90 minutes), note the timestamps of key topics for later reference.
Identify the problems: Listen through and note:
- Is the background noise consistent (HVAC) or variable (audience typing)?
- Is there room echo?
- Are there sections where the lecturer is further from the microphone?
- Are there specific loud events (chairs, dropped equipment) that need individual attention?
Processing Chain for Lecture Audio
Step 1: High-Pass Filter
Apply a high-pass filter at 100Hz. This removes HVAC subsonic rumble, HVAC vibration transmitted through the building structure, and general low-frequency noise that's below the speaker's fundamental frequency range.
Most lecturers' voices start at 100–150Hz for male voices, 150–250Hz for female voices. Everything below those fundamentals is noise.
Step 2: Noise Reduction
For HVAC and consistent background noise:
Audacity (free):
- Select a section between sentences where only background noise is present
- Effects → Noise Reduction → Get Noise Profile
- Select all → Effects → Noise Reduction
- Settings: 12–15dB reduction, sensitivity 6, smoothing 3
Adobe Podcast Enhance (free, web):
Upload the audio file and let Adobe's AI reduce the background. Excellent on voice recordings. Particularly useful for lectures where the noise reduction needs to be applied to a long file quickly.
iZotope RX De-noise (best results):
Adaptive mode tracks the varying noise levels (HVAC that cycles on and off, audience noise that varies) better than profile-based tools.
Step 3: De-reverb
For echoing lecture halls:
iZotope RX De-reverb:
- Reduction: 50–65%
- Apply conservatively — lecture halls have characteristic reverb that's part of the acoustic environment; some residual reverb sounds natural
DaVinci Resolve Voice Isolation:
For severe hall echo, Voice Isolation can separate the voice signal from the reverberant field more effectively than de-reverb alone.
Step 4: Level Correction
Lecturers move around — closer and further from the microphone. This creates significant level variation.
Compression: Ratio of 4:1 to 6:1 (higher than for podcast use because of the greater dynamic range in lectures), threshold around -20dB. The goal is to bring quiet moments up to audibility.
Noise gate: Set threshold just above the background noise level to mute background during pauses between sentences.
Volume automation: For sections where the lecturer is dramatically quieter (turned to the board, walked away from the mic), manual volume automation brings these sections up.
Step 5: De-click for Specific Events
Dropped items, microphone handling noise, chair scraping — these appear as loud transients in the waveform. Audacity's Click Removal or iZotope RX's De-click can address short events.
For very loud events that distort the recording, manual editing (cutting and cross-fading) is often cleaner than trying to process through them.
Accessibility Considerations
Lecture recordings serve students with learning differences, non-native language speakers, and people who rely on audio rather than live attendance. For these users, audio quality has higher stakes:
Intelligibility is priority: For transcription-dependent workflows (students using auto-captions, people who are deaf or hard-of-hearing), speech clarity matters more than any other quality metric. Noise reduction that improves word intelligibility, even at some cost to naturalness, serves these users better.
Consistency matters: Variable audio quality — some sections clear, others muddy — is more cognitively demanding than uniformly imperfect audio. Consistent normalization is worth the effort.
Loudness normalization for extended listening: Lecture recordings are often 60–90 minutes. Proper loudness normalization (-16 LUFS) and controlled dynamics prevent listener fatigue.
For Educational Institutions: Systematic Approach
If you're responsible for lecture recordings at an institution:
Standardize capture: Invest in consistent recording equipment for lecture spaces. A good clip-on lavalier microphone or a high-quality lectern microphone dramatically improves source quality and reduces post-production work.
Template processing chains: Create a standard processing chain in your DAW or audio editor that can be applied consistently to every lecture recording. Saves time and ensures consistency.
Automated tools: Adobe Podcast Enhance or similar AI tools can process large volumes of lectures quickly with minimal manual intervention.
Batch processing with iZotope RX: RX includes batch processing that applies the same settings to multiple files. For institutions processing dozens of lectures per week, this is essential.
Professional services for important lectures: Keynotes, high-profile guest lectures, content that will be widely distributed or archived — the investment in professional processing is justified.
WefixSound works with educational institutions and content creators on lecture recording cleanup, including bulk processing arrangements. Free sample before commitment.
Quick Settings Reference for Lecture Audio
| Processing Step | Settings |
|---|---|
| High-pass filter | 100Hz, 12–18dB/octave |
| Noise reduction | 12–15dB, adaptive |
| De-reverb | 50–65% reduction |
| Compression | 4:1 ratio, -20dB threshold |
| Noise gate | Just above noise floor |
| Loudness target | -16 LUFS (online), -23 LUFS (broadcast) |
Related articles: How to Clean Up Conference Recordings · How to Clean Up Zoom Recordings · How to Improve Voice Clarity in Audio