Audio's Circle of Confusion (2009) | Hacker News Reader

Audio's Circle of Confusion (2009)

seanolive.blogspot.com

22 pointscustomizable12 comments

The year is 2063. Audio recordings have completely eliminated all frequencies below 100hz in response to headphones' and speakers' continual drive to deliver MOAR BASE.

In response, AI-powered headphones and speakers are able to reconstruct bass notes not present in recordings by using their knowledge of the composer, musicians, and recording equipment to estimate what bass would have been there, had it not been zeroed out.

Something like this could definitely be a thing, with AI or some kind of physical simulation or both.

Imagine some Jazz recordings by some old band. There happen to be a few in high quality LP? and some in low quality (shellac?). One could imagine some kind of programming magic which transplanted only the "feel" and "timbre" of the hi quality recording onto the low quality recordings.

Further:

The simulator could build up full instruments and musicians in a studio and mike them up and modify them until the virtual simulated chain's end result is the same mix as the LP. That's a baseline. Now, you can isolate all the tracks and remix the album as if every musician was miked up separately from the beginning, even though the originally played live with a mono microphone.

It wouldn't be real, but it would be a perfectly plausable version and very close to real, too. Of course, the more you have to extrapolate and transplant, the larger the risk of ending up like that famous "enhanced" photo of Obama, but in the audio domain.

The same could be done for video, too.

Good luck getting consumers or technicians to adhere to a standardized listening environment (if that is, in fact, something that we'd even want).

I work for a consultancy that designs high-end recording studios. The single biggest factor distorting playback of audio is the room itself. And the type of room design that is required to provide an even frequency and time-domain response begins at $100k to construct. Not including our fee. For this reason I think it's out of reach for 99.9% of consumers.

We have a hard enough time convincing professional record producers and engineers of the need for a well-treated space. Many simply don't know what they don't know. And for the ones who recognize the importance of good monitoring and good acoustics many of them STILL don't know how good a room can actually be. How good it can truly sound.

There are many working professionals who've gone most of their career without experiencing a room/monitor systems that gets out of the way and allows the music the be transmitted directly to the listener. When you experience this it is visceral.

This makes it a very tough solution to sell. Even to the people who are SUPPOSED to know. So I'm pessimistic about the ability to provide that to consumers.

It seems like the experience problem could largely be solved by just having a demonstration room, but that's so obvious there must be a reason it doesn't work that way.

You're not wrong. We regularly try to get our clients into a room that we've built where they can listen. However our budget doesn't permit us to lease a commercial space and then build and retain a $400,000+ room for this purpose. Instead we often enlist past clients who will permit us to bring people into their studios to listen for a bit. They're often accommodating of us, but it's not something we can always rely on.

This article's premise seems false, or not entirely true.

The production of equipment such as speakers and microphones does not depend on recorded material.

Speakers can be tested with reference signals that did not originate in a microphone recording.

Signals from microphones can be analyzed with equipment that does not involve reproduction of those signals.

Both these transducers can be characterized by objective engineering parameters, like their frequency response curves, sensitivity, impedance and whatnot.

Obviously, to feed a signal to a microphone, we need a reference source that makes actual sound, which is some kind of speaker, and likewise to measure the output of a speaker, we need a microphone.

I don't suspect that the engineers at any respectable microphone or speaker company exclusively just sit there listening to their favorite records through the equipment, or other such subjective activities.

Just for context Sean Olive was involved in peer reviewed research to correlate objective measurements with subjective preferences in hifi. He's a leading authority in this topic area.

Ok so you define some characteristics and the room in which those characteristics were measured. Now the consumer should emulate these to standardize their listening practice? Lol.

Who is this calibration meant to serve? The audio engineers geeking out on measurements and the box they defined those measurements in? Unfortunately, 100 million people are just gonna listen to it on their airpods.

That's not the idea here. The idea is to settle on a consistent standard so that listeners don't have to constantly adjust equalization for different recordings.

I deal with this all the time. One album might sound great with both bass and treble cranked up, while another one will be waay too boom-y and I have to trim the bass down to the middle.

If some standard were agreed upon for frequency response through the chain of elements (mics, processors, monitors, etc.), then listeners would be able to set their EQs to what they like, and leave them there.

Of course I'm over-simplifiying a bit. Different artists/producers/engineers will have different subjective opinions on the "right" mix, but at least there wouldn't also be an additional complication of the variety of monitors and studio equipment settings lurking underneath their adjustments.

There is a standard. It's an acoustically treated room with monitors that produce a reasonably flat frequency response from 20Hz to 20kHz.

That's the reference.

Some items - microphones particularly - deviate from that because they can be chosen to deliberately colour the sound. But the reference itself isn't news. It's been a thing since at least the 1950s. The only difference now is that the standard is a lot more achievable and affordable than it used to be.

I doubt there's a single pro or semi-pro producer or engineer who isn't aware of this.

All but the very noobiest noobs understand that mixes need to "translate" to a variety of different speaker systems, from cheap earbuds, to car stereos, to high-end hifi.

Making a mix that does that needs good monitors, or at the very least the kind of skills and trained ear that can make a good mix on budget equipment. (Which is certainly possible, but harder to do.)

At the mid/high end it's just not an issue.

If you don't like the EQ on something, understand it's there because someone wanted it that way. It's very unlikely to be because because someone couldn't hear what they were doing.

translation is the job of a mastering engineer, not the mixer.

So what it's saying is that it's an area of space where sounds become incoherent due to interference from other sounds or reflections. So i'm assuming it affects the quality of audio recordings and the he is suggesting methods for minimizing its impact, such as using directional microphones or adjusting the distance between the microphone and the sound source. The post also touches on the importance of considering the circle of confusion in the design of recording spaces and in the placement of speakers for live events.

Audio engineers often use techniques such as microphone placement, reverberation, and equalization to control this circle of confusion and create the desired effect. For example, using a close-miking technique with a directional microphone can create a small circle of confusion and a sense of intimacy, while using a distant-miking technique with an omni-directional microphone can create a larger circle of confusion and a more spacious sound.

Reply on news.ycombinator.com