Is there a way to perform loudness normalization in Audacity?
Is there some plugin that can analyze the audio in the 1st pass, and then automatically decrease the volume of loud parts to a specified loudness level (using, for example, EBU R 128 standard), while leaving the normal parts unchanged?
I don’t know of any effect/plug-in that does that.
automatically decrease the volume of loud parts
That would be compression, limiting or automatic volume control, NOT normalization. Normalization makes ONE adjustment to the file as a whole without affecting the dynamics.
Audacity does regular peak normalization, or there is an optional RMS normalization plug-in.
I don’t know of any compressor that targets a specific loudness. You generally set the threshold (above which the compression kicks-in), the ratio (the amount of compression), and the attack & release times.
It’s a little “dangerous” to target an RMS or “loudness” level because you can end-up with clipping.
There is a Leveller effect (one of the “distortion” options) or there is an (old) stand-alone application called [u]The Levellator[/u]. But as a general rule automatic volume control often ends-up with unwanted side effects. (Neither of those use EBU R128).
There is an optional [u]ACX Check[/u] plug-in or 3rd-party plugins like [u]dpMeter 3[/u] that can give you some useful analysis/statistics on a file or a selected part of the file.
There is one app that sound like it might work. Chris’s Compressor. He designed it so he could listen to opera in the car. It works almost exactly like a broadcast compressor. No matter what you put in, it delivers a levelled show.
I change the first value, Compression Ratio from the default 0.5 to a stiffer 0.77. If I do that, it sounds exactly like a local radio station minus the radio noise.
Is there a plug-in that can normalize a few mins/seconds at a time so that there are multiple peaks at 0db instead of just one? I’m currently doing it manually in trying to increase the loudness of a conversation recording without clipping. I tried amplify but it requires clipping to be loud enough.
Sounds like you need a compressor-limiter combo plugin, like LevelSpeech2.ny
I tried amplify but it requires clipping to be loud enough.
You may be experiencing an odd audio effect. Distortion sounds loud. That’s what lets stores sell cheap music systems. If you cure the clipping the perceived volume is going to go back down again.
Are you curing the problem of recording Skype or Zoom? That’s a different problem. The far side doesn’t actually get recorded. What you got is the chat processing and echo cancellation errors. Distortion plus low volume.
Everybody On Earth is searching for a good, bulletproof way to record Chat. Nothing so far.
It’s a recording of a cassette tape. LevelSpeech2.ny IS what I was looking for. Thanks. Though it would be nice if it can normalize stereo channels independently.
While you have the stereo show in Audacity, use the drop-down menu under the black arrow on the left Split Stereo to Mono. Select each track and apply the corrections. Then Menu > Make Stereo Track.
I’ll add to this question because I don’t understand if Audacity (2.3.3) can do this, or I need to use a different application.
I have a sound file where the interviewer only used a single microphone, and left it next to him. As a result, his voice is loud… while the interviewee’s is much lower.
As a newbie, I expected Effect > Normalize to raise the low parts (and possibly lower the high part) to get an equal-souding file, but it doesn’t work.
What do you recommend I try?
Effect > Amplify and Effect > Normalize look at the whole show and change the volume once. They don’t dance down the performance changing volume every few seconds.
Did you try Chris’s Compressor? You are describing its job. Chris designed it so he could listen to opera in the car. No matter who was singing or how many instruments were playing, the show comes out even.
I use it on a spoken word podcast where the two performers are very different and it’s almost impossible to listen without constantly adjusting the volume. Doesn’t that sound like what you have?
Yes, it’s the same issue.
I’ll read up the whole thread and experiment.
Edit: Chris’ Dynamic Compressor with Compression Ratio set to 0.77 does a pretty good job. Thank you!
I call it the broadcast compressor because I used to record “Car Talk” off-air from KPCC radio. It worked, but it had background noise typical of Stereo FM radio, because it was Stereo FM radio. I could have gotten rid of the ffffffff sound by shifting the radio into forced mono, but that particular radio had no provision for that.
What I didn’t know was I was benefiting greatly from the KPCC broadcast compressors. One performer mumbled in his beer and the other had a laugh recognized as a lethal weapon by the United Nations. This came up immediately the first time I got the digital download show instead of broadcast. No compressors. It was the raw show from WBUR. It was unusable in the car.
I posted the problem on this forum and somebody brought up Chris Compressor because he had almost exactly the same problems I did. The Compression Ratio change gives a performance character almost exactly that of the station … without the background noise.
There is one bug. Chris is a look-ahead compressor and doesn’t like sailing off the end of a show. I leave some extra unimportant “stuff” on the end of a show to give Chris something to chew on and then cut off the damage later. Chris will not be fixing this because he reached end-of-life.
That and it has broadcast problems. Neither one likes long periods of silence. It throws off the broadcast averaging compressor on the hill and Chris’s algorithms.