Not -15FS anything above LSB can show up. Masking in the same critical band does not require 30 dB to mask. 0 dB will do. As the spectral density is so high from the power line harmonics more than one may fall into a critical band. So -99FS is a better goal. Keeping that out from the radiated field can be done if folks pay attention to it.
Hi John.
Agreed, sound is 'plain wrong', 'acceptable/tolerable' or 'pleasingly right', and the extension of this is how sound effects/affects one long term.
This ranges from immediately drive you out of the room, to ok/bearable through to exciting, invigorating and fun as music should be.
When a system is sounding 'right', all attention to the actual system and it's errors/faults is forgotten/ignored, and the music, performances, arrangements, production values etc become the unfettered focus and enjoyment.
Yes John, this understanding of the range of system sounds is borne of learned experience.
To the novice with mid-fi gear any small improvement in performance is seen as revelatory, to the experienced hi-ender anything less than the best experienced is sub-standard but not necessarily unlistenable.
With experience I have learned to recognise/discriminate system excess noise artifacts in reproduced sound, and this is my arbiter of good sound.
I find that for most gear, the nature of the recording embedded noise is critical, not all embedded noise is profoundly deleterious, but some embedded noises are very much so.
In lesser gear, this embedded noise modulates system intrinsic noise, and further triggers production of dynamically varying levels and spectrums of 'excess' noise, and this is where the subjective damage is done.
Ever wondered why speakers with typically rotten THD figures can still reveal differences in amplifiers/source equipment, now you know.
My filtering method enables useful reduction in production of system dynamic excess noise in every piece of gear I have tried from POS radio/cassette through to ultra hi-end...in all cases the result is transformation from crap/ok to consistently enjoyable pretty much regardless of programme.
Excess noise really is the 'deal breaker' in most systems....lousy FR, lousy THD etc don't really matter all that much in comparison.
I have the advantage that I am able to 'turn off/restore' system excess noise in immediate on the fly A/B testing....the differences are most illuminating.
Intrinsic and excess noise are more important and far reaching than generally given credit for....very interestingly I am getting playback differences in copies of the same music file transferred to a USB thumb drive.
The differences here are fine, but recognisable once heard and learned...and there is good reason for this difference I believe.
I had a quick listen to the 'Sousa' test1.
On laptop 1" speakers one file has junk that causes a subjective non programme correlated noise modulation of the main track.
I will get around to listening on better system and report back.
Dan.
I have a pair of Koss Sportapro $23 headphones, GBP 34 in the UK on Amazon (we always pay over the odds here unfortunately).
They've been making 'em for years and I bought my first pair in Japan about 10 years ago. I was blown away by the sound for such a cheap pair of cans. They are not made very well, but at $23/GBP34 who cares. If you Google it, you'll see quite a few positive reviews.
I bought them as stocking stuffers for my kids at Xmas - they loved them.
I use a pair of Audio Technica for serious listening now, but I hear modern headphones - i.e. from the last 2 or 3 years - are really good and there's been a significant step up in performance.
Amazing what a bit of competition can do.
Avoid the Dr. Dre shite is the advice I always give . . .
You didn't answer my questions. I have plenty of recordings made with even a DIY mic held in hand, hum is far down and stuff at 3k non-existant.
I thought we were talking about the PS on a DIY amplifier, so-so SNR (ITU-R 468 Curve) and say your noise dominates, -80dB re -4dBu. Why does that not put your 4k stuff at -140dB or so? Just asking a question again.
The bit was on perception in general, but particularly covered Bill Waslo's hidden band.
The power supply issue does show up in most of the low to moderate cost commercial microphone preamps I have measured. Also could be a problem in phono preamps.
Now the issue with power amplifiers is that they have massive power transformers and can put noise back on the AC mains and if foolish onto preamps very close to them.
Scott,
No idea what it is you want to ask? AC mains harmonics decrease as frequency increases. Hearing sensitivity increases just at about the same rate the harmonics decrease. Radiated noise depends on the receiving loop or system. Hearing sensitivity increases just at about the same rate the harmonics decrease. Musical masking decreases with increased frequency. The optimum bad range is 2,000 to 6,000 hertz.
Now if you have hum, the issue becomes where is it from and is it a clean 60 hertz or is it a buzz indicating harmonics. If it is a buzz then it is most likely reducing your maximum dynamic range in the mid-band. As a clean hum it does not stand out at low levels as Fletcher Munson showed. If it is a buzz then you are hearing it as that precisely because human hearing perceives the harmonics as louder than the lower frequencies. If you amplify the buzz it will round off, become smoother and sound more like a hum.
Do you have measurements of a power line hum without harmonics? Did you understand there is a threshold for perception of a tone that is influenced by in critical band noise and less so by out of band noise?
simon, you can turn up the playback level of the extracted difference track (at the bottom of the control form of the DiffMaker, labeled "playback boost") to get up to 78dB of additional gain applied. Which is of course digital gain, no new bits of information being added, just boosting the level of the hidden track during playback. I think if you do that and listen you'll not conclude that upper midrange and treble have been wiped out. In fact, other than added background hiss when the band stops the fidelity is still surprisingly good.
HF rides on the overall waveform -- just because its average energy is lower than a fullscale sine wave's, doesn't mean the HF just hangs out only in the bottom LSBs, it is a waveform going up and down with then entire waveform. Attenuation has the effect of reducing dynamic range, it doesn't do lowpass filtering!
HF rides on the overall waveform -- just because its average energy is lower than a fullscale sine wave's, doesn't mean the HF just hangs out only in the bottom LSBs, it is a waveform going up and down with then entire waveform. Attenuation has the effect of reducing dynamic range, it doesn't do lowpass filtering!
Unfortunately it doesn't want to load on my current computer. Keeps asking to set a recording device and there isn't one.
Last time I tried it, my quick look showed almost no HF on the lower level signal remnant. Not sure of the perceptual behavior of an HF signal at that low bit resolution. I'll ask a friend.
Tried a different computer. Will play a bit.
Last edited:
I asked for some very specific level numbers like what is the actual SPL of the 2-6k induced noise at the listening position vs the program content there while listening at normal levels. I guess you don't want to answer, just repeat the roll off slopes, Fletcher-Munson, etc.
Here's a plot I found in my stuff of a really bad (hum/noise) phono pre-amp that uses a 12V AC wall wart and a diode/cap voltage multiplier to make +-18V. This is playing a 1k standard reference tone. During silence the hum, etc. is audible (in headphones) but LP's don't sound too bad even though the line junk jumps above the surface noise. I would never use this regularly just an example.
Attachments
Last edited:
An interesting jump between time and frequency domain. I didn't go there, a whole new can of worms.
I listen at 60 dBa. My noise floor is 10 dBa wideband.
In your device example the AC line noise is more than 30 dB down in the critical band of the test tone and would not be perceived. However if the test tone were 57 dB lower then the noise would mask it, even though it would be perceived if the line noise were not present.
Played a bit. Getting quite different spectras from the two different pieces. Nothing much above 6,000 hertz on the recovered Sousa piece. Should be more HF from the cymbals. As you have the original can you do the analysis?
No cymbals on the choir and less to produce the higher frequencies but a flatter and fuller spectrum.
Sounds like a lot more HF in the band to me than the choir. I don't have originals anymore though. I'd not expect a lot of HF above 6kH from the cymbals of a band that isn't close up - you know how high 6kHz is, it's hardly midrange. Either way, there is plenty of HF (which attenuation doesn't filter!) and the band is clearly audible when the attenuated and extracted difference is played at the same gain as the choir (but without the choir). And the time domain behavior is way different (bouncy, transient, impacty marching band compared to smooth choir)
Last edited:
10 dBA is not wideband. Its still a remarkable achievement. I dont think Alon Wolfs Uber room is that good. You live more remote than Richard Marsh?
Sent from my LG-H811 using Tapatalk
I would have thought at that level the blood rushing through your ears and general internal peristaltic gurgles would have dominated?
- Status
- Not open for further replies.