Nada Posted January 21, 2023 Posted January 21, 2023 7 hours ago, MLXXX said: All that effort at trying to find a correlation across multiple studies, and that's all that was achieved! In general, meta-analysis can be deceptive. By including studies that had no chance through their compromised methodology of revealing a clear outcome, the process can negate excellent studies that do. In the case of RB vs HD files, better outcomes can be predicted by recruiting audiophiles who have the discriminative training to be sensitive. Pragmatically, users would be much better off testing this for themselves, eg using the protocol posted earlier in this thread. Reading about it is a very poor time investment.
MLXXX Posted January 21, 2023 Posted January 21, 2023 (edited) 1 hour ago, Nada said: By including studies that had no chance through their compromised methodology of revealing a clear outcome, the process can negate excellent studies that do. If there had been pre-existing successful studies they could have been quoted and the hi-res industry could have rested on its laurels! It was the absence of persuasive evidence from any of the previous studies considered in isolation that left no option but to try to rehash the previous investigations through meta-analysis. I recall too that some of the studies accepted into the meta-analysis were arguably dubious. I note that a compromised methodology can sometimes lead to a false positive, as opposed to a false negative. Edited January 21, 2023 by MLXXX
MLXXX Posted January 21, 2023 Posted January 21, 2023 4 hours ago, LHC said: Speaking in broad generality scientists love 'anomalies' because they could represent signals of new discoveries or lead to innovations. So when scientists encounter an anomaly in their results it is followed by a fury of effort to try and explain it away using conventional means, for example it could be a result of experimental errors, data handling/analysis errors, or calculation errors. One recent example is the B-meson decay anomaly that teased physicists of the possibility of new physics beyond standard model; after more data collection and more thorough studies this anomaly is no more and previous discrepancy attributed to errors (https://www.scientificamerican.com/article/hint-of-crack-in-standard-model-vanishes-in-lhc-data/). This is par for the course. So I naturally assumed this would be the same for Josh Reiss (now president of AES) meta-analysis finding of a statistically significant anomaly in hi res experiments, and should lead to a fury of effort in the audio research community to explain it away. Apparently this is not the case and I could only find a trickle of related research since. The general lack of interests (when compared with the broader Science norm) is puzzling. I didn't find it puzzling. In my view, the meta-analysis was a somewhat desperate effort to identify a correlation. The correlation found was extremely weak. In fact it demonstrated that for practical purposes hi-res formats were a waste of time as a distribution format (though justifiable for capturing and mixing). After that less than stellar outcome, I suggest that the last thing the hi-res industry would have wanted was a well-funded rigorous test that risked revealing that say only one in a hundred people could hear a difference, some of the time, under ideal A B listening conditions. That would have run counter to their marketing. So, best left alone, and not investigated further. 2
andyr Posted January 22, 2023 Posted January 22, 2023 (edited) On 21/01/2023 at 1:31 AM, March Audio said: It doesn't matter if the signal level is low , high or zero, the 16 bit word is recorded 44100 times a second regardless. It's a fixed amount of data regardless of what the audio signal is doing. 16 bits * 44100 * 2 channels = 1411.2 kbits per second, or 174.6 kbytes per second. And at 96kHz: On 21/01/2023 at 1:31 AM, March Audio said: 24bit 96kHz is a 24 bit word (decimal 0 to 16777215) recorded 96000 times every second. That's 4,608 kbits/s or 576 kbytes/s. Whilst I can follow your arithmetic, Alan - the figures don't match up with what I've just recorded. I just did a 'silent' rip for 3 mins (180 secs) at 24/96kHz (ie. the stylus was not in the groove). After storing the file, Windows Explorer shows the length as 136,768kB ... which is (8 bits to the Byte, right?) 1,094,144 bits - ie. 1,094.144kb ... say, 1.09Mb. This was for a stereo recording. If I take your figure of 4,608kb/s for a stereo recording ... the file would be 828Mb - which is much higher! So - given there was no input signal coming into the recording - it would appear that that the recording process is storing way lower than 4.6Mb, each second? Edited January 22, 2023 by andyr
BugPowderDust Posted January 22, 2023 Posted January 22, 2023 here’s a good calculator @andyr https://www.colincrawley.com/audio-file-size-calculator/ it backs up Alan’s numbers. As for why your files size is different, that’s a problem you need to work out.
MLXXX Posted January 22, 2023 Posted January 22, 2023 33 minutes ago, andyr said: After storing the file, Windows Explorer shows the length as 136,768kB ... which is (8 bits to the Byte, right?) 1,094,144 bits - ie. 1,094.144kb ... say, 1.09Mb. You have inadverently divided by 1000 in converting from bytes to bits. In rounded terms: 136,768kB => 1,094,144kb => 1 094Mb
andyr Posted January 22, 2023 Posted January 22, 2023 39 minutes ago, MLXXX said: You have inadverently divided by 1000 in converting from bytes to bits. In rounded terms: 136,768kB => 1,094,144kb => 1 094Mb Aah, correct - thanks M. OK, the file length is given as 1094Mb ... vs. the calculated value of 828Mb. No conflict there!
MLXXX Posted January 22, 2023 Posted January 22, 2023 (edited) 1 hour ago, andyr said: So - given there was no input signal coming into the recording There will be noise present in the lowest order bits of the ADC capture, being noise from the phono preamp whether an external preamp, or one built into an ADC* designed to be directly connected to the output of a magnetic cartridge. Possibly some low level 50Hz hum as well. If you load your recording into the freeware Audacity and apply the effect Amplify, you should be able to see the captured noise in the displayed waveform. 11 minutes ago, andyr said: OK, the file length is given as 1094Mb ... vs. the calculated value of 828Mb. What was the exact duration of your recording as displayed in file manager or on a playback device or using Audacity? You may have gone a little over 3 minutes. Edit: Also there will be overheads, see next post. _______________ * I am assuming the ADC was a 24-bit capable device. and not 16-bit. Am also assuming that the capture software did not have a squelch function to mute the signal. Edited January 22, 2023 by MLXXX
aechmea Posted January 22, 2023 Posted January 22, 2023 16 minutes ago, MLXXX said: You have inadverently divided by 1000 in converting from bytes to bits. In rounded terms: 136,768kB => 1,094,144kb => 1 094Mb Yes indeed a factor of 1000 out. And in addition ... Computer files don't just contain the raw data. There are structures imposed by the type of file. Without looking up the exact structure of a WAV file, there are likely to be header and footer records, checksums, provision for metadata and all sorts of overheads. Also there may not be an exact number of records in a block, nor blocks in a disk sector ie more waste. Back in my day we used to work on there being 10 bits per byte of raw data in our calculations to cater for overheads in a file. [One also needs to be mindful when converting from say K to M. Some people used to use the decimal 1000; others use 1024 (the power of 2 = 2**10). Maybe that has been sorted these days.] 24 x 96,000 x 2 = 4,608,000 bits/sec 4,608,000 x 180 (seconds) = 829,440,000 bits 829,440,000 x 1.25 (overheads) = 1,036,800,000 bits = 1,036.8 Mb = strikingly close to the Windows file size as calculated above by MLXXX 1
MLXXX Posted January 22, 2023 Posted January 22, 2023 (edited) 22 hours ago, Nada said: In the case of RB vs HD files, better outcomes can be predicted by recruiting audiophiles who have the discriminative training to be sensitive. Pragmatically, users would be much better off testing this for themselves, eg using the protocol posted earlier in this thread. Reading about it is a very poor time investment. As the differences between red book and higher definition recordings would primarily lie in the highest octave of human hearing, the test subjects would ideally be young (late teens or early 20s), and as you suggest have some audiophile training/experience for hearing small differences. (I used to be able to hear slight differences between different reconstruction filter selections for a DAC playing a 44.1kHz stream, but I doubt I could do that now, with my high frequency hearing now cutting out somewhere below 15kHz. Another factor, not often discussed, is what anti-aliasing filter was used to create the 44.1kHz file for distribution in the CD format, given that capture and mastering are typically done at a higher sample rate.) Unfortunately reading is all a lot of people will do. They aren't comfortable with computers and automated ABX testing which can deliver immediate A B testing and can keep a record of successful and unsuccessful trials. Instead they will take marketing claims of the audible superiority of hi-res at face value. Edited January 22, 2023 by MLXXX
andyr Posted January 22, 2023 Posted January 22, 2023 Yes, thanks @aechmeaand @MLXXX ... I'm happy that my careless error with losing the '1000' was responsible for me not getting a calculation near the actual stored value. BTW, the figure of 1094Mb is for the length of the file recorded by 'Audacity'. Just the bit stream - not converted into a WAV or FLAC file. And re. this: 37 minutes ago, MLXXX said: If you load your recording into the freeware Audacity and apply the effect Amplify, you should be able to see the captured noise in the displayed waveform. ... here is the waveform after providing 50dB of 'Amplify' to the complete 3 min file:
MLXXX Posted January 22, 2023 Posted January 22, 2023 8 minutes ago, andyr said: ... here is the waveform after providing 50dB of 'Amplify' to the complete 3 min file: Yes there was certainly noise present in that capture of "silence". If in Audacity you select "analyse => plot spectrum", you might even see a peak around 50Hz or 100Hz, being mains hum or buzz.
andyr Posted January 22, 2023 Posted January 22, 2023 4 minutes ago, MLXXX said: If in Audacity you select "analyse => plot spectrum", you might even see a peak around 50Hz or 100Hz, being mains hum or buzz. Here is my plot: How do I move the graph to the right, to extend the lowest frequ shown to, say, 50Hz?
MLXXX Posted January 22, 2023 Posted January 22, 2023 54 minutes ago, andyr said: How do I move the graph to the right, to extend the lowest frequ shown to, say, 50Hz? We're well off on a tangent here!, so I won't extend the discussion further other than to observe that that graph indicates there is some content below 100Hz (very possibly 50Hz) perhaps trending towards -14dB using the coordinates of that graph, or perhaps reaching as high as -64dB before the amplitude boosting you added of 50dB. [Given the low S/N ratio of vinyl records when played on a turntable, -64dB wouldn't be anything to worry about I'd suggest, especially at around 50Hz, a low frequency to which the human ear is not very sensitive.] * * * I'd note in passing that digital technology is remarkable in its capacity to capture and reproduce low level signals. The red book standard of "only" 16 bits actually gives more than 16 bit definition to the signal in practice because of the use of dither. The dither can be "noise shaped" taking advantage of the relative insensitivity of the ear to higher frequencies. You can do A B testing with: 1. An original 24-bit recording of music. 2. A version of that truncated to 16 bits. 3 A version of the original noise-shape dithered to 16 bits. It will be found that in quiet passages of the music, 2 may be distinguishable from 1, but 3 will not be (unless the playback gain is turned up beyond normal levels). An amazing technological achievement for audiophiles.
andyr Posted January 22, 2023 Posted January 22, 2023 Thanks, @MLXXX. What is also interesting (well, to me anyway! ) is the spike at just over 30kHz - I wonder what caused that!
bob_m_54 Posted January 22, 2023 Posted January 22, 2023 4 hours ago, andyr said: Thanks, @MLXXX. What is also interesting (well, to me anyway! ) is the spike at just over 30kHz - I wonder what caused that! If it was 20 years ago, I'd think the noise spike was induced by a VGA or XGA computer monitor close by..
andyr Posted January 22, 2023 Posted January 22, 2023 3 minutes ago, bob_m_54 said: If it was 20 years ago, I'd think the noise spike was induced by a VGA or XGA computer monitor close by.. Well, could be Bob. My monitor (connected by an HDMI cable) sits on a bench about 900m away from the PC tower case. I have no idea whether it's a VGA or XGA monitor.
bob_m_54 Posted January 22, 2023 Posted January 22, 2023 1 minute ago, andyr said: Well, could be Bob. My monitor (connected by an HDMI cable) sits on a bench about 900m away from the PC tower case. I have no idea whether it's a VGA or XGA monitor. I'm talking about the old tube type monitors, not LCD though.. the horizontal frequency for the beam. Interestingly, when you look at a lot of older music, you can notice a bit of a spike at the 15KHz from old tube type monitors.
andyr Posted January 22, 2023 Posted January 22, 2023 1 minute ago, bob_m_54 said: I'm talking about the old tube type monitors, not LCD though.. the horizontal frequency for the beam. Interestingly, when you look at a lot of older music, you can notice a bit of a spike at the 15KHz from old tube type monitors. Aah, OK - no it's certainly a flat screen. So, what can be causing that spike @ 30kHz?? Maybe the SMPS which is powering the PC?
bob_m_54 Posted January 22, 2023 Posted January 22, 2023 4 minutes ago, andyr said: Aah, OK - no it's certainly a flat screen. So, what can be causing that spike @ 30kHz?? Maybe the SMPS which is powering the PC? I think they run at much higher frequencies, ie >100KHz
andyr Posted January 22, 2023 Posted January 22, 2023 Just now, bob_m_54 said: I think they run at much higher frequencies, ie >100KHz That's what I have heard, Bob ... but I will do an experiment tomorrow. Currently the PC (with its Corsair 750w SMPS) is plugged directly into the wall. So, tomorrow ... I will put a 1KVA isolating transformer plus a 10a hash filter between the PC case and the wall socket - and re-record the 'silence' track ... and see if that 30kHz spike is still there in the signal trace.
aussievintage Posted January 22, 2023 Posted January 22, 2023 1 minute ago, andyr said: That's what I have heard, Bob ... but I will do an experiment tomorrow. Currently the PC (with its Corsair 750w SMPS) is plugged directly into the wall. So, tomorrow ... I will put a 1KVA isolating transformer plus a 10a hash filter between the PC case and the wall socket - and re-record the 'silence' track ... and see if that 30kHz spike is still there in the signal trace. Why are you going down this rabbit hole? Just a day ago you happily described this all as recorded silence ! 1
bob_m_54 Posted January 22, 2023 Posted January 22, 2023 And it may not be on the mains anyway, it may be EMI
dcathro Posted January 22, 2023 Posted January 22, 2023 I am also 16bit RB, using a tda1541 based player with no OS or digital filtering, just a simple analog filter. My own experience is that the quality of the recording is paramount, then the mastering, and finally the pressing. I avoid modern mastering and remasters, preferring the original mastering. 2
dcathro Posted January 22, 2023 Posted January 22, 2023 On 18/01/2023 at 11:14 AM, BugPowderDust said: The real reason early CDs sounded like cack was that mastering technologies had not developed alongside the media and it took some time for tools and also mastering engineers to work out how to leverage the most out of the raw media. Is this just a hypothesis, or have you done extensive comparisons? My experience is just the opposite. I seek out the earliest CDs from 1982-1986 because they sound so much better than later remasters. 2
Recommended Posts