Any way to automatically repair clipping?

I typically record my twitch stream with OBS, and then upload it to youtube. I do a bit of touching up in OBS, and I’ve found I can identify clipping fairly easily by using the Spectrogram view.

https://cdn.discordapp.com/attachments/227252967596752900/321462571230035978/5sec.png

Each white line is a clip, if I manually select the white portion of each vertical line I get fairly good results!

Unfortunately, the image above is from a 5 second clip, and I have 4 hours of audio like this to fix :confused:

is there any script that can detect these lines and use the repair tool on them? this isn’t actually peaking, it should never be above -3 db (although, for some reason on a large section I can amplify it by -4.4 db, on the clip section I can only amplify it by -1.1 db… really weird? Must be some kind of digital artifact…)

I’ve found I can identify clipping fairly easily by using the Spectrogram view.

Can you hear it? It’s not unusual to get tiny vocal sounds in overload and you can’t tell by listening.

I would forget about it. Just reduce the overall volume with Effect > Amplify: Amplification -1dB > OK

…and the red(white) lines should go away. Don’t record so hot next time.

Koz

If you’re posting an MP3, you have the additional problem of MP3 processing messing up your volume. The AudioBook people have this problem. They pass technical conformance until they make the MP3. MP3 does not like sharp, abrupt transitions and spikes.

Now that it’s too late, never use MP3 as your show master. It’s a terrible idea to edit and re-edit an MP3. The quality gets progressively worse as the compression distortion goes up—even if you save at the same quality each time.

Always Export your show master in WAV (Microsoft) 16bit. You can edit those however you want and make them into anything at any time.

Koz

I can hear it very loudly unfortunately, even if I amplify it by -10 db and so on. The only thing that makes the noise go away is using the repair tool.

This isn’t MP3, it should actually be using the CoreAudio AAC encoder at 192kbps (not sure if that’s better or worse). I would like to just record to WAV, I have the space - but it’s a pain to synch multiple audio tracks and I don’t know how to make audacity record from multiple audio devices simultaneously.

As far as I know, Audacity is a single track recorder and only will allow one input at a time. A good workflow to maintain your audio file goes something like this. Record/Import master audio file and save as “Master-Project-Name.wav”. If you add noise reduction and you like the results, save as “Master-Project-Name-NR.wav”. If you then add eq and you like the results, save as “Master-Project-Name-NR-EQ.wav” and on and on. Do this for any change you make to your master file. I promise you it is worth the extra time and will elevate the situation you are experiencing now. If someone tells you they like everything up to where it was eq’ed, then all you have to do is go back and open up your project before the eq was added and start there. Once you finish your project export to whatever your delivery method requires.

Audacity can record multiple “tracks” at the same time, but only from a single “device”. If a device has multiple channels and the device driver presents that to applications as one multi-channel device, then you can set the number of channels in the device toolbar to match the number of available channels and record those channels simultaneously. In the case of exactly 2 channels, Audacity records as a two channel stereo track.

DeClicker plugin script for Audacity here.
The processing-time with the DeClicker plugin is at least playback-time,
(that plugin may choke on hours of audio, try 15 minute chunks at first, then increase the length of the audio to see what it will cope with on your computer).