Ending the Windows Audio Quality Debate

[antcollinet]

A distortion peak measured at sum or difference of two prime numbers played from separate stereo systems cannot be derived from any of them independently, thus it can only be the microphone distortion. Non prime numbers will not be sufficiently conclusive.

Isn't that just proving that a microphone distorts? Guess what : so do your ears.

Perhaps link to the video, or paper or whatever. Then we can all understand better what the claim is.

 

[Andro]

Isn't that just proving that a microphone distorts? Guess what : so do your ears.

Perhaps link to the video, or paper or whatever. Then we can all understand better what the claim is.

 

[Andro]

 

[AnalogSteph]

This unfortunately is quite common in inexpensive measurement mics (and small electret capsules in general). At high levels, they distort, and when moved further away, their noise floor becomes a problem. The Linkwitz mod is a thing for a reason. My old one was so bad that I resorted to using a side address LDC.

 

[Andro]

At any rate, I no longer care about measurement microphones, I have nothing left to measure. As I said here:

Ending the Windows Audio Quality Debate

@edechamps you're right about volume control, I forgot some DACs have a hardware volume control. My understanding from the past (it could be outdated) that most affordable DACs (under $1k) have either no hardware volume volume control or they are some low quality Chinese stuff. If you want both...

audiosciencereview.com

at this stage of the system fidelity (thanks to the last link of EAPO) if I hear any distortion I know it's in the recording (like in the example of Dire Straits) and not in my system. I have only to wait for the iconic music gems of the past to be finally remastered in HD.

 

[dasdoing]

it should be pretty obvious that these cheapo mics can't measure loudspeaker distortion accurately. You just have to measure a loudspeaker at different distances while maintaining its volume the same. the distortion gets higher the nearer you go. no complicated rocket science is needed.
but they are more than enough for room EQ

 

[Sokel]

it should be pretty obvious that these cheapo mics can't measure loudspeaker distortion accurately. You just have to measure a loudspeaker at different distances while maintaining its volume the same. the distortion gets higher the nearer you go. no complicated rocket science is needed.
but they are more than enough for room EQ

For distortion measurements REW says the same,these typical mics are no good (also USB such as UMIC1 cannot be used because of unstable clocks but UMIC2 can be used)
However it does suggest that for FSAF measurements an Line Audio Design Omni1 or a AKG P170 can be used but only if calibrated for FR against an already calibrated for FR mic.

 

[dasdoing]

For distortion measurements REW says the same,these typical mics are no good (also USB such as UMIC1 cannot be used because of unstable clocks but UMIC2 can be used)
However it does suggest that for FSAF measurements an Line Audio Design Omni1 or a AKG P170 can be used but only if calibrated for FR against an already calibrated for FR mic.

has anybody ever checked the chips inside those USB mics? they have a ADC and a pre inside that tiny space. how good can they be?

 

[Doodski]

has anybody ever checked the chips inside those USB mics? they have a ADC and a pre inside that tiny space. how good can they be?

Good enough to take a calibration and come out pretty accurate for a mic. SMD devices can take up such small spaces and afford such amazing space constraints design(s).

 

[AnalogSteph]

It's usually some sort of single-chip USB codec from the usual suspects (Realtek, CMedia, perhaps even a TI oldie). The converters and filters tend to be decent enough, and even preamp noise is generally adequate for an electret mic. I would consider the ability to perform external filtering of the mic bias voltage essential though, as a noisy mic bias voltage is a very common electret mic performance killer when used with onboard audio and the like.

 

[Andro]

After few weeks of playing with Equalizer APO my conclusion is that it's indeed great when fixing a sloppy crossover work, however fixing the passive crossover and removing the need of equalization in exclusive mode sounds far superior to EAPO. After my last passive crossover redesign it's almost flat:

If I try to improve it with APO it actually sounds worse. So:
Good crossover in exclusive mode >> Sloppy crossover with EAPO >> Sloppy crossover without EAPO

A soldering iron and a few good capacitors will do you more good than Equalizer APO.

 

[AnalogSteph]

It goes without saying that every EQ is only ever as good as the measurements it's based on (MMM works a treat IME, even if you only have a 0° mic cal file, that just makes the measurement a bit more cumbersome), and it can't fix actual dispersion issues or poorly suppressed breakup modes. (Obligatory endorsement of woofer series notches goes here.) You also need to keep adequate headroom for it (PEACE is good for keeping an eye on levels).

I would definitely want to address the substantial low-midrange dip (a single low-Q filter should do it) as well as the utter lack of much of anything below 100 Hz (the 50-100 Hz range is super important, and I'd want decent coverage to ~60 Hz at least).

I just came back from measuring some Panasonic SB-PM01 micro stereo speakers from the very early 2000s, which had recently been liberated from being stuck behind the monitor for years and then even found their way onto DIY desk stands (fancy!). Despite some reference level tweaking, 13 bands wouldn't do it, so I ultimately resorted to throwing 17 bands at the automatic EQ and later adding an 18th for manual bass boost sub-100 Hz... these are little 4" jobbies from a quarter century ago, not that much going on down there by default. They sound like a million bucks now - well, proverbially anyway, but I was definitely quite pleased with the result.
Only the BH3857AFV electronic volume in the SA-PM01 CD receiver unit is clearly not up to dynamic classical, it's getting quite noisy if you crank it. (Hey, it's an old micro stereo that was bought to accompany an Athlon 1200 system if memory serves, or at least moved to PC duties fairly quickly. These never were intended to be "real" hi-fi.)

Hey look, somebody has even gone to the trouble of developing a crossover mod for these speakers. Yeah, the stock response is quite... lumpy, so you can see why I needed a gazillion bands. Honestly, given the single capacitor XO I'm surprised they sound as good as they do now... yay for direct sound in nearfield, I guess. (The midwoofer also seems to be quite well-behaved, with severe breakup only at 8-9 kHz.) This might still make a decent project. Unfortunately I have zero crossover parts laying around so would have to buy everything. At this point I'd rather want to sort out the hissy amplifier situation first.

 

[Andro]

@AnalogSteph 1. So true. In fact a parallel notch filter is my main weapon to greatly improve the speaker clarity and reduce distortions. This is why audiophiles say that DSP and active crossover speakers are inferior to the passive crossovers - there is nothing that can kill cone breakup resonances than a good 14 gauge inductor in series with the driver. Any driver is an electro-mechanical device and increasing its inductance dampens the target resonance frequency.

2. Very good advice, I've bumped the sub-100 range and it sounds much fuller, so EAPO is a keeper. I am still listening if it adds any mid and high distortions but nothing audible yet.

 

[Andro]

Here someone did some measurements of Spotify and found TPs above 0 dBFS.
Streaming services encode lossless masters in lossy codec and this results in a 99% increase in TP (see iZotope).

No they are not. All streaming services out there except Spotify are lossless, they do it to compete with Spotify, which is a complete piece of garbage but is still popular because most people don't care. You get worst case CD quality but right now Tidal already has over 10% of tracks (over 10 mln) in Hi-Res and is adding more literally every week. Just stop using Spo***trash.

 

[Jimbob54]

All streaming services out there except Spotify are lossless,

This clearly isn't true.

 

[Davide]

No they are not. All streaming services out there except Spotify are lossless, they do it to compete with Spotify, which is a complete piece of garbage but is still popular because most people don't care. You get worst case CD quality but right now Tidal already has over 10% of tracks (over 10 mln) in Hi-Res and is adding more literally every week. Just stop using Spo***trash.

Even platforms that make lossless their flagship offer lossy streaming, because it is obviously useful and convenient in some scenarios, such as mobile devices. And this is often handled automatically by the respective apps.
The point of the discussion was the possible introduction of intersample overs in the conversion to lossy.
To consider Spotify trash just because it is not lossless is to ignore two things, namely that in 99% of cases a human cannot distinguish lossy from lossless and that Spotify's preference algorithms are like tailor-made clothes; or at least better tailored than the competition. Without forgetting that lossless, in some cases, is a significant additional price.
Fortunately, there is competition anyway.

 

[CedarX]

To consider Spotify trash just because it is not lossless is to ignore two things…

Agreed, but I would still consider Spotify trash because it ranks toward the bottom in royalties payment to the Artist.

 

[meleeli]

Why disable "system sounds" btw? No actual justification is given for this in the first post, can't find anything searching in this thread.

 

[Jimbob54]

Why disable "system sounds" btw? No actual justification is given for this in the first post, can't find anything searching in this thread.

So you dont get notification bleeps etc on top of the music

 

[Sokel]

Why disable "system sounds" btw? No actual justification is given for this in the first post, can't find anything searching in this thread.

On top of what @Jimbob54 writes, system sounds often are way more louder than what you listen too, sometimes close to 0dB if you haven't adjust them specifically.
You can imagine the havoc they may create.