Videos don't do well on Hacker News, but I encourage people to at least watch the first couple minutes of this one. The oscilloscope visual overlay is interesting and the editing is really good.
Also, given the topic (audio equalizers) there's no way it could have been a blog post.
I would hope mostly what doesn't do well is useless titles - like one word, or a pithy joke that makes sense only in retrospect. Unfortunately there is also that guidelines which discourages doing better.
I think the title on HN is misleading.
Sadly I’m not sure what’s the short and clearest title :(
The video is very nicely made and it focus on sound systems / boomboxes frequency response and behavior of included filtering modes.
So when talking about EQs with “all forms” in mind, you should consider:
- EQ is merely combination of one or more filters.
- There are many common filter designs, the video isn’t about that, it also doesn’t mentioned Low-pass/High-pass/band-pass/bell or common structures, but it is only showing them.
- Filters made and behave differently
- Most filters (including the ones in the video) are done on the time domain (vs spectral one)
- Phase, this is the biggest missing piece in the video imho. Naive filtering is “smearing” the signal to achieve the different tone balance. By doing so, they also most likely change the phase (unless using linear phase filters)
You have a slight misunderstanding here about the phase shift bit.
Technically filters don't cause phase shift, phase shift causes filters.
That means in the analog world all realtime filters come with phase shift, if they didn't you wouldn't have a filter.
In the digital world there is a thing called linear-phase mode where clever people found ways of building filters that shift the phase of the whole signal the same way (involves running the signal through forwarwards and backwards through special filters). If you now have an audio program that deals correctly with the latency introduced by each effect, you can shift the rest by the same amount meaning you bought yourself a linear phase response with added latency (a small delay) on everything. Remember: Humans can't really hear phase smears unless they happen in relation to another thing that is in phase. This only happens when you mix two signals. The result of that mix is yet again a filter (e.g. comb filters).
With filters everything is a trade off all the time, but again phase shift is not a just a bad side effect of a filter, it is the mechanism by which it does the work. And sometimes the phase shift introduced cannot be heard.
And also linear phase isn't for free, besides the latency, it also adds pre-ringing, which especially in the lower frequencies can be very audible and annoying.
For the earlier systems with the bass boosts, are the graphs showing what comes out of the cables or what comes out of the speakers?
I would assume they don't care about the very low bass because they were designed as a unit and physically couldn't produce those notes very well anyway. Hence the need to boost slightly higher frequencies to fake bass.
The straighter lines are what comes out of cables.
The more squiggly ones later (eg: 2:45) are what comes out of a speaker. But since he uses a microphone to measure the speakers, so that measurement will also come with room sound. He would need an anechoic chamber to measure "only the speaker".
In the 70s/80s every home stereo system -- racks of stereo equipment stacked a meter high -- had a dedicated equalizer. It was not just for audiophiles!
I suspect HiFi culture was slightly different in different parts of the world.
In my part of the world, the original HiFi guys rejected the EQ units. I was influenced by that I think (my father was an original). I still have a metre high stack of HiFi, in almost daily use, and have never felt the need to have an EQ unit.
There was a company called DAK in the 80's that sold all sorts of interesting stuff and I still use my BSR EQ-3000 which is an equalizer with a spectrum analyzer display fed by a microphone that you walk around the room with to confirm your settings. It even has a pink noise button that injects that into the amp so you have a uniform pattern to equalize. Sort of like an analog Sonos Trueplay I guess. We could do all of this in the 80s :-)
I built my own 10 band equaliser using instructions from Elektor (I think) back in the 80s. The dream was to acquire a spectrum analyser but real life intervened.
Not sure when I even powered up my stereo system. Probably doesn't even work now.
Couldn't hear this video, just watch, because I have no sound in the env I am currently in.
But some thoughts about EQ.
"EQ isn't phase shift, phase shift is EQ.", a video title I saw once, but hits the nail on the head.
"Nothing is for free.", if you eq some frequencies, you affect some other frequencies too. You can use this to your advantage, if you are aware of this.
"EQ graphs are lying." Lack of resolution and not revealing the phase shift makes most stock EQ graphs not telling the truth.
"My hearing aids sound strange/metallic." This is the phase shift, you are hearing.
A pleasant, if muted, layman's 10 minute journey to explain sound, eq, and speaker response. It gets juiciest on room eq correction/compensation, approaching the 9 minute mark.
Pink noise distributes energy logarithmically in a way that matches the the sensitivities of the human ear. The graph which he shows on screen is ALSO in that same logarithmic space, hence why it shows pink noise as flat: it's flat as far as how humans hear it. White noise is not perceptually flat to humans. Thus when testing audio, it's important to use pink noise instead of white noise.
More often than not, yeah. On the music production (rather than listening) side the two EQs and spectrum analyzers I use most frequently, FabFilter Pro-Q and Voxengo SPAN, both use a 4.5dB/octave tilt by default. You can adjust it, but I've never done so.
I tried to post Posy's videos a few times here, but with no bigger interest.
His videos about LCD technology are hypnotizing. Bonus - he makes all the music.
His video always scratch a part of my brain that i event dont know it exists.
Videos don't do well on Hacker News, but I encourage people to at least watch the first couple minutes of this one. The oscilloscope visual overlay is interesting and the editing is really good.
Also, given the topic (audio equalizers) there's no way it could have been a blog post.
It's pretty common for blog posts in this arena to just include samples that you click to play.
Videos are harder to watch at work since it's obvious they're not work related compared to blog posts.
Good point. The bandcamp link to Posy music is equally pleasant. Nice to find actual mellow music.
I would hope mostly what doesn't do well is useless titles - like one word, or a pithy joke that makes sense only in retrospect. Unfortunately there is also that guidelines which discourages doing better.
Usually the subtitle is fine:
> A video about all forms of equalizers. From one-click bass buttons to advanced studio correction.
That can be reduced to:
"Equalizers: From bass buttons to advanced studio correction."
PS: I'd try to keep the original title too, but in this case it doesn't look nice
"EQ: Equalizers - From bass buttons to advanced studio correction."
I think the title on HN is misleading. Sadly I’m not sure what’s the short and clearest title :(
The video is very nicely made and it focus on sound systems / boomboxes frequency response and behavior of included filtering modes.
So when talking about EQs with “all forms” in mind, you should consider:
- EQ is merely combination of one or more filters.
- There are many common filter designs, the video isn’t about that, it also doesn’t mentioned Low-pass/High-pass/band-pass/bell or common structures, but it is only showing them.
- Filters made and behave differently
- Most filters (including the ones in the video) are done on the time domain (vs spectral one)
- Phase, this is the biggest missing piece in the video imho. Naive filtering is “smearing” the signal to achieve the different tone balance. By doing so, they also most likely change the phase (unless using linear phase filters)
- Filtering might result delay in time.
You have a slight misunderstanding here about the phase shift bit.
Technically filters don't cause phase shift, phase shift causes filters.
That means in the analog world all realtime filters come with phase shift, if they didn't you wouldn't have a filter.
In the digital world there is a thing called linear-phase mode where clever people found ways of building filters that shift the phase of the whole signal the same way (involves running the signal through forwarwards and backwards through special filters). If you now have an audio program that deals correctly with the latency introduced by each effect, you can shift the rest by the same amount meaning you bought yourself a linear phase response with added latency (a small delay) on everything. Remember: Humans can't really hear phase smears unless they happen in relation to another thing that is in phase. This only happens when you mix two signals. The result of that mix is yet again a filter (e.g. comb filters).
With filters everything is a trade off all the time, but again phase shift is not a just a bad side effect of a filter, it is the mechanism by which it does the work. And sometimes the phase shift introduced cannot be heard.
And also linear phase isn't for free, besides the latency, it also adds pre-ringing, which especially in the lower frequencies can be very audible and annoying.
For the earlier systems with the bass boosts, are the graphs showing what comes out of the cables or what comes out of the speakers?
I would assume they don't care about the very low bass because they were designed as a unit and physically couldn't produce those notes very well anyway. Hence the need to boost slightly higher frequencies to fake bass.
The straighter lines are what comes out of cables.
The more squiggly ones later (eg: 2:45) are what comes out of a speaker. But since he uses a microphone to measure the speakers, so that measurement will also come with room sound. He would need an anechoic chamber to measure "only the speaker".
In the 70s/80s every home stereo system -- racks of stereo equipment stacked a meter high -- had a dedicated equalizer. It was not just for audiophiles!
I suspect HiFi culture was slightly different in different parts of the world.
In my part of the world, the original HiFi guys rejected the EQ units. I was influenced by that I think (my father was an original). I still have a metre high stack of HiFi, in almost daily use, and have never felt the need to have an EQ unit.
There was a company called DAK in the 80's that sold all sorts of interesting stuff and I still use my BSR EQ-3000 which is an equalizer with a spectrum analyzer display fed by a microphone that you walk around the room with to confirm your settings. It even has a pink noise button that injects that into the amp so you have a uniform pattern to equalize. Sort of like an analog Sonos Trueplay I guess. We could do all of this in the 80s :-)
I built my own 10 band equaliser using instructions from Elektor (I think) back in the 80s. The dream was to acquire a spectrum analyser but real life intervened.
Not sure when I even powered up my stereo system. Probably doesn't even work now.
I love Posy's channel so much. Really beautiful love letters to all kinds of vintage tech (plus his own music!).
The history behind vU meters is also fascinating. People used to calibrate off the BBC techs, or off American techs.
No BIPM, no SI units: what the BBC say.
I knew this was a Posy video before even opening the link.
Couldn't hear this video, just watch, because I have no sound in the env I am currently in. But some thoughts about EQ.
"EQ isn't phase shift, phase shift is EQ.", a video title I saw once, but hits the nail on the head.
"Nothing is for free.", if you eq some frequencies, you affect some other frequencies too. You can use this to your advantage, if you are aware of this.
"EQ graphs are lying." Lack of resolution and not revealing the phase shift makes most stock EQ graphs not telling the truth.
"My hearing aids sound strange/metallic." This is the phase shift, you are hearing.
Have a nice Sunday everybody!
A pleasant, if muted, layman's 10 minute journey to explain sound, eq, and speaker response. It gets juiciest on room eq correction/compensation, approaching the 9 minute mark.
I'm not sure Posy is a layman.
Afaik the dude is a professional audio engineer with a retro soundsystem hoardin.. ahem, "collecting", hobby.
He's decided to collect no more than what fits into the white cabin.
This is a fantastic yt channel!
Highly recommend their lightbulb video as well.
This is my favorite: https://youtu.be/eGQQWIbD-nM?si=CtQ-MPIxx8h3vYxH
Posy is just one guy.
Although he has at least two personalities. Sorry. Sorry.
Very first sentence: "This is pink noise... if you measure it, you can make it look like a straight line."
He then shows a horizontal straight line: that's white noise?
Pink noise distributes energy logarithmically in a way that matches the the sensitivities of the human ear. The graph which he shows on screen is ALSO in that same logarithmic space, hence why it shows pink noise as flat: it's flat as far as how humans hear it. White noise is not perceptually flat to humans. Thus when testing audio, it's important to use pink noise instead of white noise.
https://en.wikipedia.org/wiki/Colors_of_noise#Pink_noise
> The graph which he shows on screen is ALSO in that same logarithmic space, hence why it shows pink noise as flat
Does this also mean that audio equipment in general will display a spectrum with this kind of logarithmic offset adjustment?
More often than not, yeah. On the music production (rather than listening) side the two EQs and spectrum analyzers I use most frequently, FabFilter Pro-Q and Voxengo SPAN, both use a 4.5dB/octave tilt by default. You can adjust it, but I've never done so.
Fffff