UnDistort Audio File
UnDistort Audio File is a powerful audio file repair tool, to fix pops, cracks, glitches and fix silent skips in audio files.
It does not remove humm or hiss for example, only sharp changes in the audio.
UnDistort Audio File is a shareware software. Unregistered version is a fully functional trial version that adds beeps in saved audio files. If you like this software please register.
- Automatically detect distortions in audio files with adjustable detection parameters
- Display each detected glitch one-by-one visually on a sample display
- Fix all distortions automatically or manually one-by-one
- Apply the frequencies that are present before the detected distortion to the fix
- Save the fixed audio file in MP3, Opus, Flac or WAV format
- Multi-channel audio files are supported
To fix the detected distortions 2 ways are possible:
- Select an input file and click "Load".
- Click "Run auto calibration", the "Treshold" specifies how many distortions to detect minimally.
- Click "Find distortions". If there are too many detections, adjust the "Detecting options" parameters and click "Find distortions" again.
- Click "Next distortion" (multiple times) and check the detection results.
- To fix all detected distortions automatically click "Fix all detected distortions" (highly not recommended, as false detections will actually spoil the audio).
- Go through all the detected distortions one-by-one and either fix or remove the detected distortion.
To fine tune the particular distortion fixes use the "start" position, "end" position, "Waveform steepness scan length" and "FFT modulation" parameters until the suggested displayed fix seems ok.
If removing them, after removing all not needed detected distortions, click "Fix all detected distortions", this will apply the fix for all the distortions that were not removed.
Click "Save" to save the new audio file with the fixed distortions.
- Or if clicking on "Fix this distortion" for all the distortions one-by-one (recommended), if finished, click "Save" to save the new audio file with the manually fixed distortions.
- The audio samples are only modified if clicking on "Fix all detected distortions" or on "Fix this distortion".
Distortions fixer detection parameters
Use the sensitivity slider to quickly change the 2 most important parameters for detection. 100% is usable for audio files with average volume.
Set "Treshold" to specify how many distortions to detect minimally when using "Run auto calibration".
When loading an audio file "Run auto calibration" should be the first thing to do.
If there are only a couple of clicks for example in the audio, set "Treshold" to little above the number of clicks.
To fine tune the detection the parameters are as follows:
- Check sample difference: specifies the sample difference of one sample after the next, that triggers a check whether it is a sharp change occuring.
- Forward check sample count: specifies how many samples to check after the trigger, to decide whether it is a sharp change really.
- Forward check sample sum difference: specifies the average sum of the "Forward check sample count" samples' values.
If the calculated average is higher than this value a detection start point is detected.
- Check sample difference continue: specifies the maximal difference of the first detected sample's value, where the distortion started, and the sample's value that's the last sample value of the distorted part (where the distortions ends).
So if the sample value is in this range it is considered as an ending point of the distortion.
- Max. distortion length: maximal distortion size in samples to treat as a detected distortion.
Detected distortions that are longer will not be treated as a detected distortion and will be skipped.
Skip silences fixer detection parameters
Skip silences fixer detection works on the currently selected channel.
- Sample maximal value: max. value of samples where after silence detection starts to scan.
- Minimal silence length: minimal length of a silence part. If detected silence part is shorter it will not be detected as a silence part.
- Maximal silence length: maximal length of a silence part. If detected silence part is longer it will not be detected as a silence part.
- Max. silence samples values: when deciding if the detected silence part is valid, values that are above this value will be counted against the "Silence treshold".
- Silence treshold: when deciding if the detected silence part is valid, values count that are above this % the part will be considered as not a skip.
The 2 most important parameters are the Minimal silence length and Maximal silence length.
Fine tuning these 2 parameters helps the most with detection.
Using the cut out mode though will cut out the audio on all channels, to keep the channels in sync.
The reconstruct mode is applied only to the selected channel. So to use this mode all channels need to be scanned and fixed individually.
Using the cut out mode should fix most skips on other channels as well implicitly, as skips are usually occuring at the same position on all channels at once.
If using the "Fix all detected distortions" functionality all fixes will use the "Waveform steepness scan length" and "FFT modulation" values.
- Waveform steepness scan length: number of samples to average when calculating the steepness of the start and end of the distortion. The suggested fix's shape is determined from the calculated steepness. Higher values result a more average fix shape, lower this value to have a flatter fix.
- Start smpl and End smpl: if the detection is not precise, use these values to adjust the detection's start and end point. Only usable when doing manual fixing.
- FFT modulation: 256 sample values are processed just before the starting point and their frequency distribution is calculated with an FFT, the slider specifies how much to apply these frequencies to the fix.
If fixing the detected distortions one-by-one manually (recommended) these values can be changed to apply the fix every time with specific values.
Important to precisely adjust the start and end position of the fix until it visually looks good.
- It is advised to use WAV audio files as input.
Convert the source files to WAV with eg. Context Menu Audio Converter.
- The input audio files should not be longer than 1 hour.
- Perform multiple passes, go with the parameters for a reasonable distortion count first, eg. 10.
If around 10 distortions are found, go through these detections and fix them.
Save the file, restart the app. (just for sure), load the saved file and try to set detection parameters to find again 10 distortions.
Again fix them, save the file again, and repeat these steps, until needed.
- When fixing skips, first go through the first channel, fix the skips as needed (save the file), then switch to next channel, do a scan and fix the skips on this channel.
Repeat this for all channels, then try to set parameters more relaxed and do a scan and fix again.
Repeat these steps, until needed.