JP
7480 74111110101115
My unit has a variable resistor for impedance load. I dialed in the sound.
Which also changes the level. You should check out the test files (if/when the site is back up).
Effects of MC Loading
- Thread starter JP
- Start date
Which also changes the level. You should check out the test files (if/when the site is back up).
There was thread here on AK where a forum member had posted a plot of an Audio Technica LOMC cartridge (can't remember which model) at a load of 100R and 47k. The plots were identical except for a slight shift in level. Unfortunately, I doubt I'd be able to find it again.
D.D.P.
New Member
This post is from a couple of years ago but was brought up again a couple of months ago at SHF. Perhaps this is what you were thinking of?
AT OC9 xsh settings
I have a couple of the Audio-Technica OC9 cartridges here. Have the XSH and the XSL, with the XSH being on my SOTA Cosmos Eclipse/SME V. Hello Mr. Chancho! Thanx for sharing! Beautiful table!! :bowdown:I am sure you get the most out of these AT. What are the audible differences between the XSH...
audiokarma.org
JP
7480 74111110101115
With my phono pre-amp and the Denon DL-103R and DL-103SA, 200-ohm sounds better (fuller) than 100-ohms.
I use headphones to dial in the crazy thing.
Ya, that’s the kind of stuff that happens when you don’t match levels.
Yes, that's it - well done on finding it.
JP
7480 74111110101115
I'm probably going to steal that script of yours in the future for my own abx testing
GitHub - jaakkopasanen/ABX: Web app for AB and ABX listening tests
Web app for AB and ABX listening tests. Contribute to jaakkopasanen/ABX development by creating an account on GitHub.
github.com
AvFan
Super Member
I'm want to better understand some of the concepts regarding human hearing that are the basis for the listening tests. I found this NIH paper with lots of good information but I found this statement particularly interesting with regards to this discussion:
So considering a threshold of 0.1dB or even 0.5dB seems very conservative. But I find it fascinating that a 0.5% shift in frequency can be heard. And that brings me to the process to level match the samples as described by @mr_petit with @JP 's concurrence:
What exactly is Audacity doing in its loudness normalization? Could that process have any effect on the frequencies?
Sound Discrimination
Over a range of frequencies (approximately 500 to 4000 Hz) and levels (approximately 35 to 80 dB SPL) in which humans are most sensitive, listeners can discriminate a change of about one decibel in sound level and about a half of a percent change in tonal frequency. For instance, a 50 dB SPL sound can be just discriminated from a 51 dB SPL sound, and a 2000 Hz tone can be just discriminated from a 2010 Hz tone.So considering a threshold of 0.1dB or even 0.5dB seems very conservative. But I find it fascinating that a 0.5% shift in frequency can be heard. And that brings me to the process to level match the samples as described by @mr_petit with @JP 's concurrence:
What exactly is Audacity doing in its loudness normalization? Could that process have any effect on the frequencies?
mr_petit
High on Vinyl
No, loudness normalization is purly a gain/reduction adjustment over the complete spectrum
Loudness Normalization | Audacity Support
Apply this normalization effect to set the target loudness required by podcast platforms, television/radio programmes and some websites
support.audacityteam.org
You can play around with a sigal generator yourself and experience the small differences you can distinguish.
It totally depends on the situation. If you switch back and forth, you can easily distinguish differences much smaller than 1dB, but if you go and get groceries inbetween and listen again, you couldn't distinguish 10dB of difference.
So if the setup of the test is like: "if you think the next tone is louder than the previous tone.....press 1, If you think the next tone is quieter than the previous tone....press 2......beep", then yes, that 1dB is probably very correct. Acoustic memory is very short.
It may also depend on the type of signal. Are we talking here about sine tones, or a more complex signal like a piece of music or a conversation?
Last edited:
AvFan
Super Member
No, loudness normalization is purly a gain/reduction adjustment over the complete spectrum
Loudness Normalization | Audacity Support
Apply this normalization effect to set the target loudness required by podcast platforms, television/radio programmes and some websites
You can play around with a sigal generator yourself and experience the small differences you can distinguish.
It totally depends on the situation. If you switch back and forth, you can easily distinguish differences much smaller than 1dB, but if you go and get groceries inbetween and listen again, you couldn't distinguish 10dB of difference.
So if the setup of the test is like: "if you think the next tone is louder than the previous tone.....press 1, If you think the next tone is quieter than the previous tone....press 2......beep", then yes, that 1dB is probably very correct. Acoustic memory is very short.
Thanks for the link. I didn't see it noted but do you think the sample rate for the loudness normalization is the same as file's sample rate?
mr_petit
High on Vinyl
I don't know exactly what you mean by this.
It calculates percieved loudness from the samples selected and then adjusts it to the desired value.
The adjustment is a simple gain adjustment over the complete selected audio. There's no change of samplerate.
The gain adjustment is the same for every sample, every frequency.
How it handles stereo or mono tracks and what the options are, are explained in the link.
Last edited:
AvFan
Super Member
I could be way off on all this but again from the NIH paper:
The thresholds for detecting a tonal sound increase as the duration of the sound to be detected decreases at durations shorter than 500 ms, but remain approximately constant as the duration increases above 500 ms.
The conclusion of all this is that a change in loudness is perceived as tonal differences when making granular changes in MC loading but if the loudness is constant then different MC loadings can't be discerned. If Audacity is adjusting the gain using 1/2 second intervals of data to determine the adjustment could it impact the perceived frequencies? If Audacity is adjusting the gain at the sample rate of the digital file then there would be no issue. I'm wondering if Audacity's algorithm to make the loudness change might inadvertently "homogenize" the perceived frequencies in the sample recordings resulting in the conclusion that loading does not matter.
lini
just me...
Btw, there's a convenient online test, whether you actually can discriminate +/- 1 dB differences - see there: https://www.audiocheck.net/blindtests_level.php?lvl=1 At my notebook I can only very reliable discriminate +1 dB, listening via its integrated speakers and at comparatively low volume, though - whereas telling flat from -1 dB is more like guesswork for me. Unfortunately the next easier step, that online test offers, is for +/- 3 dB, which is too easy.
Well and regarding those 0.5 % in frequency, I've checked that with this online tone generator: https://www.szynalski.com/tone-generator/ Result: Yup, 0.5 % work for me - in either direction.
Greetings from Munich!
Manfred / lini
JP
7480 74111110101115
This is worded a bit ambiguously - it means the discrimination of level, not any discrimination at all.