A Perceptually Meaningful Audio Visualizer (2016) | Hacker News Reader

A Perceptually Meaningful Audio Visualizer (2016)

delu.medium.com

111 pointsChadNauseam33 comments

I love the fact that TFA essentially re-created the functionality of an oscilloscope. Except, much more interesting to look at, and clearly shows an understanding of FFT math that I sometimes think I understand but then read stuff like this and remember I don't really.

Here's an example of using the O'scope: https://www.youtube.com/watch?v=19jv0HM92kw

You can make an affordable sound visualizer out of a tea cup, used speaker, led light and duct tape. Attach the tea cup to the speaker, add some water, drop a submersible led light and turn on the speaker. The vibration will produce patterns on the water that will refract the light onto the ceiling. With a decent quality setup, you'll see a mesmerizing quality projection of a 3d shape. This trick is used by dolphins to encode visuals in sound, and I've seen a speculation that thoughts in your imagination have the same nature.

I think i prefer spectrograms. Particularly for speech, where you can, after a while, almost read them.

I would love a small desktop spectrogram waterfall that shows the output from whatever I'm listening to at the time

I did a few CTF challenges where I wouldn't have even thought to check the audio until I heard some uncharacteristic noise. Wonder what else people have buried...

https://github.com/vsergeev/audioprism

this looks about like the thing..

If you're on Windows, I've seen a few spectrogram skins for Rainmeter that uses whatever audio is being played on the device. For example, there is AudioAnalyzer[0]. Personally, I've been using a skin that shows a simple waveform of the current system audio, but also integrates with Spotify to change the color to match the primary album color. Unfortunately, I don't have the name or link on hand.

0: https://forum.rainmeter.net/viewtopic.php?t=31091

oh sweet! https://forum.rainmeter.net/viewtopic.php?f=27&t=34430#p1701... did the trick :) ty

Run a soundcard loopback into an instance of izotope insight (or similar).

I have this exact same idea on my audio synthesis and DSP exploration project called Akasha, that I have developed on and off for more than 13 years. I have plans to open source at least parts of it, once I get to clean up the code. I was inspired by Julius O. Smith III's writings on DSP algorithms: https://ccrma.stanford.edu/~jos/

Unfortunately my web pages are not online, I have some video samples of Björk, Gotan Project and some classical and electronic music samples.

In the analytical audio signal, the musical intervals can sometimes be easily seen, and it is mesmerizing to look at.

On this video demo by OP, when the waveform appears stationary with loops, there are intervals in (at least close to) just intonation – meaning the frequencies have integer ratio relationships.

https://en.wikipedia.org/wiki/Just_intonation

the code is open source. I did a but of work a few months back updating it and fixing compilation with rust stable toolchain: https://github.com/khimaros/audioscope

I wish visualization could be more expressive like this.

Nightingale & Canary - Bird sounds visualized by Andy Thomas

https://www.youtube.com/watch?v=NtIHJSRQvvk

WOW

(and I searched for more and found https://andythomas.com.au/pages/bird-sounds)

That’s incredible

I find this to be a lot more informative, though: https://0x0.st/oFP6.png

That is a snippet of birdsong from that video, plotted on time/frequency axes so you can see the structure of the call. Color denotes sound intensity at a given (time, frequency) region.

I wonder if the author had ever tried milkdrop before? Seems like the same type of thing. https://en.m.wikipedia.org/wiki/MilkDrop

Milkdrop is an open source reimplementation of the old-school WinAmp stuff, so it tends to have the same disappointing visualizations that don't match well.

Disappointing? What?? milkdrop is decades old, still works (while AVS crashes on startup) and has very good presets (altough editing is a world of wizardry).

It's not really synesthetic, but, nonetheless, what DID you expect?

I found this mesmerizing. It conveys surprisingly many of the perceptual characteristics of the sound through visuals. Good demo music, too, but it makes me curious to see what it would do with vocals or acoustic music. Any other demos?

Not sure about the colors… they were hard to make sense of. And I wonder if it’s making use of stereo information at all.

I'd also like to know how it visualizes other music. The song seems like it was crafted specifically for the demo, or at least I think that sometimes it's too good with the spirals sitting still - but it may as well just be the combination of this brilliant visualization and the sine waves of whatever signal.

Hey folks, I wrote this article! If you're interested in a simpler reference implementation, I made it in Web Audio: https://github.com/conundrumer/visual-music-workshop/blob/7b...

I've also continued developing new audio visualizers, like this one inspired by sands on a vibrating plate: https://twitter.com/conundrumer/status/1482615130185768961

Interesting, and I'd love to see some more demos on "traditional" music (Symphonic/Orchestral, Jazz, Beatles, etc.). I'm sure the the highly synthetic and "pure" music used in the demo video sure leads to a fairly coherent visualization as compared to more natural/acoustic sources of sound.

Also, what about any stereo components? Capturing that in the visualization would also be nice.

Possibly worth mentioning that this is from 2016.

So cool! Recreated this in typescript:

https://dcyoung.github.io/r3f-audio-visualizer/?mode=AUDIO_S...

https://simplespectrogram.app/

Gives me the following error on an iPhone:

“Sorry, there was an error: The request is not allowed by the user agent or the platform in the current context, possibly because the user denied permission.”

Ooh, intriguing.

I've been using Flux studio session analyser to visualise audio signals for mixing In-ear monitors for live audio[1] from my Cue mix.

Might write a quick dodgy thing to add this to my analysis stack. Could be cool.

[1] https://imgur.com/RFsV3d9

There's a Google Experiment from 2018 using Hilbert Scopes:

https://experiments.withgoogle.com/seeing-music

Does anyone know of any chanting apps to give visual feedback on your voice as a tool for meditation?

Demo here: https://vimeo.com/196216785?embedded=true&source=vimeo_logo&...

If one is interested in audio visualisations, you should check out Cycling74's Max MSP with Jitter. That program has huge capabilities in making these things!

Would like to feed this voice samples, cat's meows, birdsongs, and laughter. Maybe the same voice in different emotional states. Looks fun to play around with.

Love this. Many hardware oscilloscopes can (and were) configured to do this in polar plotting mode.

Not the same, although the resulting shspes might look similar, a Lissajous pattern displays both channels of a stereo signal as X/Y. That means independent of what is going on if you send it a mono signal, you will just see a 45 degree line — totally independent of the frequencies contained in the audio.

This uses FFT to draw circles and makes the harmonics/frequencies visible.

Need winamp plug-ins!

Reply on news.ycombinator.com