What Audio Problems Cannot Be Fixed (And What Can)
One of the most important things to understand before using an audio restoration service — or spending hours on DIY processing — is what's actually fixable and what isn't.
Audio professionals deal with disappointed clients who were told (or assumed) that restoration could achieve more than physics allows. This guide gives you honest, specific expectations so you know what you're working with.
The Core Principle: You Can Only Enhance What Was Recorded
Audio restoration works by analyzing what's in the recording and improving it. It cannot add information that was never captured. This sounds obvious, but its implications are widely misunderstood:
- It can remove noise that was mixed in with a voice signal — because the voice signal is also there to preserve
- It cannot restore frequency content that a microphone never captured — because there's nothing to restore
- It can reduce reverb — but it can't make a recording sound like it was made in a different room than it was
- It can reduce distortion from mild clipping — but it cannot recover a signal that was completely overloaded
With that principle in mind, here's a systematic guide to what's fixable and what isn't.
What Can Be Fixed: High Success Rate
Consistent Background Noise
AC/HVAC hum, computer fan noise, refrigerator buzz, consistent traffic — these have predictable, consistent character that noise reduction algorithms model well. A recording with a steady background noise can typically be cleaned to the point where the noise is inaudible or barely perceptible.
Success rate: Excellent. This is the most reliably solvable audio problem.
Electrical Hum (50/60Hz)
Interference from power systems. De-hum processing applies notch filters at the fundamental and harmonics and removes it cleanly without affecting the voice signal.
Success rate: Excellent. One of the easiest audio problems to fix.
Vinyl Crackle and Clicks
Surface noise from vinyl records. De-click tools identify the impulsive events and remove them. Light to moderate surface noise can be reduced to nearly inaudible levels.
Success rate: Very good for mild-moderate. Heavy groove damage becomes more difficult.
Tape Hiss
The characteristic noise floor of analog tape recordings. Adaptive noise reduction reduces it effectively. The target is "not distracting" rather than "completely absent."
Success rate: Very good with appropriate tools. Be realistic that some residual tape character is expected.
Mild Room Echo (Small Untreated Room)
A bedroom or home office recording. De-reverb processing at moderate settings significantly reduces the hollow quality.
Success rate: Good. The recording will sound noticeably more controlled.
Mild Digital Clipping
Brief peaks that occasionally exceeded 0 dBFS. De-clip processing reconstructs the peak shapes.
Success rate: Good for mild clipping. Degrades with severity.
What Can Be Partially Improved: Moderate Success Rate
Moderate Room Echo (Larger Reverberant Space)
A conference room, classroom, or large home with significant echo. De-reverb processing reduces the echo substantially, but some residual reverb remains at higher reduction levels.
Success rate: Moderate. Results are noticeably better but not fully clean.
Variable Background Noise
Traffic that varies in intensity, audience noise in a lecture hall, wind noise outdoors. Adaptive tools track the varying level but can't model noise that changes character.
Success rate: Moderate. Reduction is uneven; some sections better than others.
Internet Compression Artifacts (Zoom, Phone)
The "robotic" quality of compressed calls. AI-based voice separation tools can partially reduce this, but the underlying compression artifacts are baked in.
Success rate: Moderate. The recording sounds better but not like a clean local recording.
Moderate Digital Clipping
Clipping that occurs more frequently but isn't continuous. Recovery is partial — harsh moments are improved but not fully clean.
Success rate: Moderate. Each clipped section improves independently based on how severe it is.
Recording at Significant Distance from Microphone
A room microphone picking up a voice across a large room. The ratio of direct-to-reverberant sound is unfavorable. Processing helps, but the fundamental problem of low signal-to-noise ratio limits results.
Success rate: Moderate. Can be improved but won't sound like a close-miked recording.
What Cannot Be Fixed: Low/Zero Success Rate
Severe Digital Clipping
When the majority of the signal was recorded above 0 dBFS, the signal is fundamentally compromised. The flat-topped waveform contains destructive harmonic distortion, and there's too little intact waveform data for reconstruction algorithms to work with.
Why it can't be fixed: The original signal information in the clipped peaks is gone. There's nothing to recover.
Missing Audio / Extended Dropout
Sections where the audio simply isn't there — tape dropout from oxide shed, digital corruption, recording interruption. If there's silence or unintelligible noise where audio should be, that content is gone.
Interpolation can fill very short gaps (under 500ms) with estimated audio from surrounding content. This works for brief clicks or dropout. For longer gaps, the content isn't recoverable.
Missing Frequency Content
If a microphone or recording system never captured frequencies above a certain point — a very cheap microphone with 4kHz rolloff, a phone call with 3.4kHz bandwidth, a recording through a wall — no EQ can add that frequency content. Boosting a frequency range that contains only noise amplifies noise.
Why it can't be fixed: You can't enhance what was never recorded.
Simultaneous Overlapping Voices on Single Track
If two people spoke at the same time on the same microphone track, they cannot be separated. The acoustic superposition of the two voices is mixed into the signal — no algorithm can reliably separate them because the signals occupy the same frequencies at the same times.
Why it can't be fixed: The separation problem requires information about each voice independently, which wasn't captured.
Severe Reverb from Extreme Environments
A recording made in a cathedral, large industrial space, or swimming pool may have a reverb time of 3–10 seconds. De-reverb processing helps at lower settings, but aggressive reduction creates severe artifacts before the reverb is eliminated.
Partial improvement is possible but the recording will not sound like a treated space. Manage expectations accordingly.
Audio Altered or Corrupted at the Codec Level
Some types of corruption — truncated files, codec errors, data corruption — produce audio that has lost its digital integrity. Depending on the specific corruption, restoration may be possible or impossible. File repair tools (not audio restoration tools) address some types of corruption.
Psychoacoustic Masking of Critical Information
If background noise or music is louder than the voice you want to hear, and both are present in the same frequency range at the same time, the wanted signal may be masked below audibility. You can reduce the noise, but if the voice was recorded at a level below the noise, it may not become audible — or may only be partially intelligible.
How to Evaluate Your Recording Before Submitting
Before investing in professional restoration, evaluate your recording honestly:
Can you understand what's being said? If the content is completely unintelligible due to noise, the signal-to-noise ratio may be too poor for meaningful improvement. If you can understand the content with effort, restoration can make it effortless.
What's the dominant problem? Noise, echo, clipping — identify the primary issue. Match it against the categories above.
Is the problem consistent or variable? Consistent problems respond better.
How much of the recording is affected? Occasional problems in an otherwise clean recording are easier than problems throughout.
The Free Sample Answer
The most reliable way to know what's possible with your specific recording: WefixSound provides a free 60-second sample. Submit the worst section of your recording — the part you most need fixed — and see exactly what restoration achieves before committing to payment.
This removes all the guesswork from the "is this fixable?" question.
Related articles: Audio Restoration Service: What to Expect · How Much Does Audio Restoration Cost? · AI Audio Restoration vs Human Engineers