PulseAudio Loudness Equalizer for dynamic contexts such as movies | Hacker News Reader

PulseAudio Loudness Equalizer for dynamic contexts such as movies

github.com

99 pointsrendx34 comments

I just installed pulseeffects today because I wanted to have a noise-gate and compressor on my microphone for work (whereas I use lv2 plugins with JACK on my personal machine).

The "app" and "mic" both showed up as outputs and I couldn't figure out how to make zoom use the "pulse effects mic" and then I realized it was just haphazardly hijacking the actual mic rather than letting me pick it as a separate output. And pulseeffects needs to stay in a window running to work, with no minimize to tray option that I could tell?

Felt useful if you only needed it running sometimes in certain scenarios, but felt like it didnt fit my use case, or I was definitely missing something.

> And pulseeffects needs to stay in a window running to work, with no minimize to tray option that I could tell?

You need to launch PulseEffects as a daemon ahead-of-time, then if the daemon is running the GUI will attach to it. You can launch the daemon by running `pulseeffects --gapplication-service`. In the GUI, if you go upper-right-hamburger-menu→General, there's a toggle for "Start Service at Login", which will write a ~/.config/autostart/pulseeffects-service.desktop file that runs that command. Most desktop environments will do the correct thing with that.

PE (I've used it only for listening, not recording) takes a while to 'sink in' (pun intended). At least some of the PE preset settings still function when it's 'closed'. In ?Ubuntu it'll minimize with the panel icon visible (if you've include it there). Some options can't be enabled without installing 'lsp-plugins'. 'pavucontrol' is essential to see the goz-intas/outas.

In the lower left of the Zoom conference window is a mute button. If you look closely, you'll see an arrow there, which, when clicked, causes a drop down menu to appear, where you can pick your audio input and output devices.

TYL! Today You Learned!

yeah... except both "app" and "mic" pulseeffect devices appear under "output devices" and there is no additional input device... it appears that the standard mic input is being modified without being a second device, so I'm not 100% sure what the other devices are for, if it just in-place modifies existing devices.

have you try NoiseToch ? it's a dumb to use AI noise filter (like the NVIDIA filter) that works over pulse audio. It's very impressive.

Just tried as per your recommendation, I have a loud AC unit behind me but the high noise floor is entirely gone. It's almost magical, sure it sounds a bit like low bitrate lossy compression artifacts but it's incredibly clear.

I'll be using this for all my calls in the future.

> have you try NoiseToch¹?

¹ NoiseTorch, not NoiseToch.[0]

[0] https://github.com/lawl/NoiseTorch

A somewhat lighter-weight approach with PulseAudio (vs running full-blown PulseEffects) can be achieved through the use of ladspa-sink plugins, as described briefly on the Arch wiki for dynamic range compression https://wiki.archlinux.org/index.php/PulseAudio#Steve_Harris...

Using "plugin=sc4_1882 label=sc4 control=1,1.5,401,-30,20,5,12", I am generally able to smooth things out pretty well.

I don't understand why anyone would voluntarily use compression on anything. I'm very sensitive to these kinds of compressor and can literally hear the gain ramping up and down.

If you want movies to have lower dynamic range for some reason, use the built-in dynamic range compression that comes with Dolby Digital.

If you want equal loudness between different songs or albums, use replaygain. My entire library has replaygain applied and its wonderful to be able to keep my stereo set at a decent level and just press play on anything. Without it you always have to turn the volume knob down first "just in case".

aaah can the youtube player please add loudness equalization? so many lectures have terrible sound interspersed with horrifying loud noises.

it's a good service to users, but frankly demanding it of each app separately is folly & ruin. computing should be more useful, you should be able to depend on your OS to provide you basic services like this.

ditto for tools like virtual background.

coupling our services to the consuming apps is a big mistake, leading to users facing inconsistent offerings & each app needing it's own configuration wizards. this serves no one & is not how competing within computing should happen, in siloes.

I needed this so badly in my life. I use mpv even for music this is a godsend.

I'm unfamilar with this tech, but I think I need this badly in my life. Care to explain a little about what you'd use this for?

I think I need this because I listen over headphones to audio that changes volume levels (eg. Spotify playlists with a variety of artists) in a somewhat noisy office environment (I can't just listen more carefully when the music is quiet) but don't want to turn the volume up too high. I also don't want to have to be constantly reaching for a volume slider. Currently, I am constantly reaching for a volume slider, because some song comes on that's loud and I have to turn it down, then a quiet song comes on and I forget to turn it back up, and eventually I notice I'm distractedly listening to conversations from sales instead of focusing on my tasks and drowning them out with background music. And heaven help you if you have a source of audio with ads (eg. Pandora or Youtube playlists), where every half hour you'll get your eardrums blown out with some obnoxiously loud ad break.

My uneducated expectation is that this would run in the background on my Linux laptop, letting the volume decay in between beats and as the song switches, but amplifying quiet songs and compressing loud ones. It would be nice if some of the dynamic range of a song was preserved, but instead all songs should go from, to use the musical terms, mezzo-pianno to mezzo-forte. I don't particularly care if it has to buffer the next 10 seconds of the song to make that work, but as I like to stream a variety I can't wait for it to compress a full hour of audio.

Is that what this does? Do I need to use mpv, or have a local/offline mp3? Is there a better way to do what I want?

Didn't read the article but I'm almost certain it's dynamic range compression.

I don't know why this hasn't been standard in Every OS.

EDIT: NOPE! It's just an automatic gain control :/

Dynamic range compression is kind of the same thing as automatic gain control (well, automatic gain control is one way of describing how to implement dyanmic range compression). This github project implements a compressor, it says in the readme.

Spotify includes loudness correction in it’s preferences. It will compare the LUFS loudness of all songs and turn down the loud ones to match -14 LUFS.

It works as expected on my phone, although I notice it is off on my desktop app, despite it being activated in my preferences. A bug.

Sennheiser headphones with ALC might help - they do a pretty good job for me in evening out volume levels.

I think you might be able to accomplish this in mpv using --lavfi-complex and the apprpriate ffmpeg filters.. alimiter maybe?

since my update to ubuntu 20.4, I have to kill pulseaudio regularly :-(

I’ve never found one that works as well as the one included with Winamp in the late 90s. Or have I just become fussier since then?

Through cumulative experiences interspersed with informed analysis, you've become more refined and more discriminating in your tastes, and simultaneously more capable of articulating your preferences with sophisticated nuance.

These tendencies towards a more informed, articulate aesthetic are coloured by a parallel inclination to recalibrate past experiences in terms of your newfound aesthetic epistemology, whilst integrating an elevated enthusiasm from an epoch in which your tastes were less refined and your enthusiasm for musical novelty was more pronounced.

Also Justin Frankel is a good programmer. Ease up on the thesaurus, buddy :)

Aka Dynamic Range Compression, right?

I really wish we could agree on standard terms for these things...

It contains a dynamic range compressor, but it also chains more effects including a multiband compressor, so overall it falls outside the definition of a dynamic range compressor. Naming chains of effects is always an interesting exercise, I prefer using non-existing words to avoid confusion.

Not the same thing. Normalization is about scaling the volume to either the peaks or a weighted average.

It doesn’t change dynamic range, just absolute value.

Dynamic range compression is something that happens on relatively short timescales, like say, 10ms-1 second.

This does not perform normalization. The Github repo does have the tag "audio-normalization" but the tags seem a bit all over the place.

From the readme

>An Upward Compressor is used to raise low level signals ...

Is there any difference between a 'loudness equalizer' and a 'compressor'?

Kind of. A compressor is usually a dumb dynamics processor that just looks at an envelope to compute gain reduction.

A loudness equalizer is a much smarter dynamics processor that can have gain reduction or addition and tracks perceptual loudness (not the envelope) while adapting parameters to hit some target loudness.

Its sort of like a really smart AGC, which is a general purpose dynamics processor.

Possibly working on specific frequencies and boosting/cutting those in addition to compressing transients. Sometimes referred to as “Multiband compression”

I just use the VLC audio compressor.

I tried with vlc years ago and wasn't impressed with results. Maybe it has gotten better. But I had thought, wouldn't a process be needed that would prescan the entire file to determine the highs and lows of the audio?

You'd need to scan the whole file if you were trying to normalize the volume, but, generally real-time audio compression is pretty good (and what you want) if configured well.

You're almost never interested in whether the audio you're listening to is absolutely the loudest or softest across the entire file, you typically just want to minimize sudden changes that leave you straining to hear something unexpectedly quiet or bracing against something unexpectedly loud. This compressor works in two directions - downward compression on loud parts, and upward compression on soft parts, bringing them both closer to the median loudness - but preserving some smaller level of dynamic range, so things still sound "good".

Reply on news.ycombinator.com