High-frequency restore
Filters out high-frequency noise and hiss first, then applies standard restoration. Worth trying first on recordings with obvious hiss or electrical whine.
Local audio restoration for Mac and Windows: repair recordings with hiss, noise, and missing frequency ranges
Repair damaged voice recordings
Some recordings have more than a bit of noise: poor equipment, low bitrates, cut-off highs, heavy hiss and noise floor make them sound muffled and muddy. PureRestore uses an AI model to rebuild the missing and damaged parts of these recordings so the voice becomes clear again. It offers three restoration modes and handles single files or whole folders. The model installs with the app, all processing happens on your computer, and no network is needed.
Feature set
Balanced restore
The default mode: standard restoration, suitable for everyday work on most recordings.
Deep restore
Uses a different analysis method for restoration. Results are consistent between runs but can sound quite different from the standard mode. For badly damaged recordings where the first two modes aren't enough.
Single file or a whole folder
Pick one file or a folder. While processing, the stage, progress, and remaining time are shown for each file, and you can cancel mid-run. Failed files state the reason, such as insufficient disk space or a decode error.
Runs fully on your machine
The restoration model ships inside the installer: no downloads, no network. Recordings and results are never uploaded.
Command line
Restore in batches from the terminal, with input, output, restoration mode, and overwrite or skip options. Settings made in the interface can be copied straight into a command.
In depth
How it works
- Pick input: a single audio file or a folder (batch processing).
- Pick a restoration mode: high-frequency restore, balanced restore, or deep restore; balanced is the default.
- Start restoring: the progress page shows the current stage and estimated remaining time.
- Check the results: the result page lists files found, succeeded, failed, skipped, and the elapsed time. Results are saved as new files; originals are not modified.
You can run the same recording through different modes and keep the version that sounds most natural.
Which mode to pick
| Mode | What it does | Try it first when |
|---|---|---|
| High-frequency restore | Filters high-frequency noise and hiss, then applies standard restoration | Obvious hiss or electrical whine |
| Balanced restore | Standard restoration | Most recordings; start here |
| Deep restore | Uses a different analysis method; the result may sound quite different | Heavy damage, the first two aren’t enough |
Restoration is reconstruction, not recovery. The model fills in what it judges to be missing; the heavier the damage, the further the result may drift from the original sound. Keep the source file for important recordings.
How it differs from PureRefine
- PureRestore repairs damaged recordings: missing highs, muffled sound, heavy hiss and noise floor.
- PureRefine cleans intact recordings made in a bad environment: background noise, room echo.
For recordings with both problems, repair with PureRestore first, then remove the remaining noise and echo with PureRefine.
Command line
You can run PureRestore in a terminal and put restoration into batch scripts:
--input: a file or a folder--output: output location; each input file gets its own result folder--mode fast_restore | balanced_restore | deep_restore: high-frequency, balanced, and deep restore respectively; default is balanced--overwrite/--skip-existing: overwrite or skip existing results
The Mac version runs in a sandbox, so the CLI can only reach directories authorized in settings; the Windows version accesses the paths you give directly.
Privacy, performance, and requirements
The restoration model installs with the app, about 660 MB. Processing never goes online and recordings are never uploaded. PureRestore has no account system and no usage statistics. The network is only used when you open a web link yourself.
- Platforms: macOS and Windows, paid downloads from the Mac App Store and Microsoft Store, buy once with no in-app purchases.
- Input formats: WAV, MP3, FLAC, M4A, AAC, OGG.
- Output format: WAV.
- Mac: computes on the CPU by default; longer recordings take proportionally longer.
- Windows: NVIDIA CUDA first, then DirectML, falling back to CPU when neither is available.
- Interface languages: Chinese, English.
Who it is for
- Podcast and video creators: narration and interviews recorded on phones, cheap mics, or in noisy places
- People digitizing old recordings: tape transfers, low-bitrate voice memos, repeatedly compressed audio
- Online classes and meeting recordings: calls recorded remotely, muffled sound with electrical hum
- Anyone processing in bulk: put the restoration step into automated scripts via the command line
FAQ
Does PureRestore upload my recordings?
No. The restoration model installs with the app and processing happens on your computer; no network is needed.
How is it priced?
A paid download, buy once. After purchase everything is included, including folder batch processing and the command line. No in-app purchases and no subscription.
Which mode should I try first?
Start with the default balanced restore. When hiss or electrical whine is obvious, try high-frequency restore; when damage is severe and results aren't enough, try deep restore. You can run all three on the same recording and compare.
Does the restored audio sound the same as the original?
Not exactly. The model fills in what it judges to be missing; the heavier the damage, the bigger the possible difference. Keep the source file for important recordings; results are saved as new files.
How is it different from PureRefine?
PureRestore repairs damaged recordings, such as missing highs, muffled sound, and heavy hiss. PureRefine removes background noise and room echo from intact recordings. You can use them in sequence.
How long does one file take?
It depends on the recording length and your computer's performance. The Mac version computes on the CPU by default, so long recordings take longer; on Windows it is much faster with an NVIDIA GPU.
Which formats are supported?
Input: WAV, MP3, FLAC, M4A, AAC, OGG. Output: WAV.
Tech and privacy note
PureRestore is designed for offline local processing and does not depend on cloud transcoding. Read ourprivacy policyfor full data-handling details.
Support and feedback
You can reach us via email feedback@efficient-lab.com or submit issues and suggestions through GitHub Issues.
Related products
PureLab
Local music and sound-effect asset manager for Mac and Windows: AI tagging, batch organizing, safe writeback
PureSpleeter
Local stem separation for Mac and Windows: split a song into vocals, accompaniment, drums, bass, and more
PureRefine
Local voice denoising and de-reverb for Mac and Windows
PureDSD Batcher
Batch-extract a whole folder of SACD ISOs into tagged DSF files