John Curl's Blowtorch preamplifier part III

Status
Not open for further replies.
Here's a picture of the attack green original blue with removal of everything over 22050. The Gibbs is seen at the new fs/2, the subtraction is done on a full 10sec clip at once (1,920,000 points) so the separation of the spectra shows only the numerical noise floor (~-250dB). The horizontal scale is in samples (5.2us) the vertical scale is arbitrary, I had to convert the 24bit into 32bit (Python does not do odd bytes) and then process at 64bit FP no loss of resolution or truncation happened AFAIK.

I guess everybody understands why I find this great example ueber funny :rofl:.
 
It was then.
In the mid-seventies, I said to an Ampex guy: "16 bit is not enough, I want 24 bit". He laughed and said: "I'd love to sell you a 24 bit converter, it'll be about that size." and pointed to a full 19" rack.
And so 16 bit was cast in stone, and the dark age begun.

That's a fairly narrow view and would end up with nothing to show for 100's of millions of dollars and thousands of man years of research. If someone said in 1980 we can't release a digital audio product unless it does 24 bits at 96kHz there would be no digital audio.

Rudy van der Plassche (the designer (or head of team) of the original TDA1541) visited us in 1982 and made the comment that the upper management at Phillips had no idea if the original CD format would succeed and were prepared to write the whole thing off if need be.

Remember RCA did write off their entire CED video effort.
 
Here's a picture of the attack green original blue with removal of everything over 22050. The Gibbs is seen at the new fs/2, the subtraction is done on a full 10sec clip at once (1,920,000 points) so the separation of the spectra shows only the numerical noise floor (~-250dB). The horizontal scale is in samples (5.2us) the vertical scale is arbitrary, I had to convert the 24bit into 32bit (Python does not do odd bytes) and then process at 64bit FP no loss of resolution or truncation happened AFAIK.

Quite like expected, right? :)
There will be a silence here, because you had to listen ;)
 
Disabled Account
Joined 2012
IMMIC, Careful researchers like F.Toole wrote that we can detect amplitude changes of 0.1db in broadband signal (music).

I dont know if a Blurred sound is caused by the amplitude difference. Doubt it. But difference here is significant IMO.

We have not yet found why some do hear differences with CD compared to live (and of course some dont). Some people think that Bose player sound real. Bose said so, too. Cant use that argument as too many do hear differences still when playing their CD. .



THx-RNMarsh
 
Last edited:
That is so cool, thanks.
What does the difference look like in the same time period?
Odd the spectra difference is minimum 250db down,yet the differences are clearly there. I see places where the green is almost 20% higher peak than the blue, with the same basic waveform.

Maybe I was not clear, the difference of the spectra has only something above 22050 and numerical noise below 22050 which is the same as in the time domain the spectrum of the difference only has data above 22050.
 
Maybe I was not clear, the difference of the spectra has only something above 22050 and numerical noise below 22050 which is the same as in the time domain the spectrum of the difference only has data above 22050.
Nah, you were clear. However, the green peaks are at many times significantly higher than the blue ones. They don't quite look like higher spectra, but same.

What is the difference between the two waveforms you provided? It looks like it will have quite a lot of energy within the lobes at the same frequency as the blue.

If you don't want to try that, can you send me a copy of the data on that graph that I can plop into excel?

jn
 
Last edited:
Am I misreading it or does the filtered version actually have a faster rise time?

Not really I think it shows that the transients don't stress the 44.1kHz sampling. What I don't have is a good way to make a 44.1kHz file that translates directly to this 192kHz file even though there is exactly no content above 22.05K I have no access to any hardware capable of doing the 44.1 to 192 conversion at this level of precision.

Again theory and implementation are at odds.
 
Member
Joined 2002
Paid Member
Using a general purpose math tool like Python/numpy I can make a perfect brickwall filter by simply zeroing all the bins past 22050 (this is a real only FFT) and get the expected perfect sine wave taking the inverse FFT.

Scott, is it possible to upload a wav (44.1kHz/16bit) file of the 20kHz sine brickwall filtered past 22050 Hz?
Thanks

George
 
Close.

Some remarks. I don't (as many of my colleagues) don't agree with Ethan Winer advocating Omni-directional mics as a requisite. We all use cardioid mics, for good reasons, and close miking, for good reasons. Just they need an EQ to compensate for the proximity effect.
A the question of "where" the music was playing, the answer is "in your room".
Close miking is the only way I know to get a realistic recording of an instrument. If you want to mimic distance, reverberation has to be added. If you do long distance miking, you'll get, may-be, a correct distance feeling, but you will lose details and sound quality of the recorded instruments that your ears hear in real time. Note that, in any case, to get the right distance feeling, the mic has to be set closer than your listening position of the real event.

+1 from another recording engineer. And heres why: Critical Distance calculator

An omni mic will pick up as much reverb as direct sound at as little as 2 feet from the source meaning you will hear as much of the room as the instrument and you loose the details. ( all highly dependent on the room, the instrument, and the mic)
 
And does your repo system sound as accurate as live compared with CD with music playing?

If so, you are done. Enjoy. If not, does 24/192 sound closer to reality?

THx-RNMarsh
There is no way I can get the acoustics of a concert hall into my music room, not even close.
I don’t even try to compare, it’s a different world.
The only thing that’s important to me is that I can enjoy the music, with perfect articulation of voices and violins where you can also hear the woodsounds instead of just the strings
And most important of them all, that you don’t get tired of listening after 15minutes.
Most of the time I listen to LP’s because not all CD’s are recorded with TLC, nothing having to do with sample frequency.

Hans
 
If you don't want to try that, can you send me a copy of the data on that graph that I can plop into excel?

jn

No problem, but jeez I'm going to have to re-figure out how to do Dropbox. You might have some fun with Python, it's a one click download with everything you need. All this stuff is about 10 or 20 lines of code.

In the meantime I'll figure out how to put that data into a file but it will be big.
 
Disabled Account
Joined 2012
Backing up a bit.... so we know the filter is may be Ok enough for 20Khz/44.1 etc. maybe.

JC said and many others... musicians, reviewers, studios, customers it was the Tr. Or related to.

Others add that CD doesnt do the highs well, it gets congested or blurred.

Are we off the track again as to what is needed to sound more accurate? ?

What other ways are there to improve on CD sound??

Is it on the record side? Compounded rec/play??


Why does 24/192 et al sound more accurate to a lot of people? Myself included.



THx-RNMarsh
 
Last edited:
Disabled Account
Joined 2012
There is no way I can get the acoustics of a concert hall into my music room, not even close.
I don’t even try to compare, it’s a different world.
The only thing that’s important to me is that I can enjoy the music, with perfect articulation of voices and violins where you can also hear the woodsounds instead of just the strings
And most important of them all, that you don’t get tired of listening after 15minutes.
Most of the time I listen to LP’s because not all CD’s are recorded with TLC, nothing having to do with sample frequency.

Hans

of course not. i am talking about the sound heard in direct/near field sound only. Like in near-field or headphones. There you dont have much or any room. It is a prefered way to listen if you appreciate accuracy.

You might not want to always listen that way but try it to learn how close to real your repo system is without sever room influence.

You might not care. Thats fine.


THx-RNMarsh
 
Last edited:
Scott, is it possible to upload a wav (44.1kHz/16bit) file of the 20kHz sine brickwall filtered past 22050 Hz?
Thanks

George

OK, but I'm not sure what you are asking for. A 20K sine wave at 44.1kHz looks like the modulated versions shown there is nothing to brickwall filter as the input, 20kHz, has no content above 22.05kHz. The interesting thing is the 20kHz sine at 44.1kHz upsampled to 352.8kHz and then brickwall filtered where all the modulation is gone.
 
Status
Not open for further replies.