Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: DC sample for LocrianGroove tests (Read 8801 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

DC sample for LocrianGroove tests

Here is a 14s sample with pre-emphasis and DC in the file. LocrianGroove wants to test around with it.
Is troll-adiposity coming from feederism?
With 24bit music you can listen to silence much louder!

DC sample for LocrianGroove tests

Reply #1 –
Quote from: Wombat on
Here is a 14s sample with pre-emphasis and DC in the file. LocrianGroove wants to test around with it.
This is a bit of a random guess - but I don't think that file was really recorded with pre-emphasis. The high frequency noise floor of the recording is visible in parts of it. That "noise floor" is far above -90dB (so is due to the instruments, not the ADC / conversion), and most importantly, that noise floor is spectrally flat.

If the signal on the CD was encoded with pre-emphasis, you'd expect the noise floor to raise at higher frequencies, like this...
http://www.audioxpress.com/magsdirx/ax/add...ia/galo3025.pdf
(see figure 1).

There's no such rise, so I doubt this signal is really recorded with pre-emphasis.

Maybe it's a mistake, or maybe the artist recorded it flat, but then (bizarrely) decided they preferred the sound with the de-emphasis applied?

Cheers,
David.

P.S. I wouldn't worry about the DC offset either. At least in that example, it's less than 1%, and could be entirely attributable to the asymmetric waveform of some of the synth notes, rather than an actual DC offset. If you're looking at other parts of the file, you might see things that demonstrate a real DC offset that are hidden here.

P.P.S. - unless the pre-emph and DC offset were magically removed in fb2k when I converted the flac to wav!
 

DC sample for LocrianGroove tests

Reply #2 –
EAC say FLAGS PRE. Soundwise it is hard to tell. Sounds still ok with de-emphasis since it is only synth music.

Edit:
I verified with icedax:
defaults   stereo, 16 bit, 44100.00 Hz, track 1, no offset, one track,
          type wav 'audio', don't wait for signal, not quiet,
          use generic_scsi, device /dev/cdrom, aux /dev/cdrom
    @    :~$ icedax -D /dev/scd0
Type: ROM, Vendor '_NEC    ' Model 'DVD_RW ND-4550A ' Revision '1.09' MMC+CDDA
569344 bytes buffer memory requested, 4 buffers, 55 sectors
#icedax version 1.1.9, real time sched., soundcard, libparanoia support
AUDIOtrack pre-emphasis  copy-permitted tracktype channels
      1- 2          yes              no    audio    2
AUDIOtrack pre-emphasis  copy-permitted tracktype channels
      3- 3          yes            yes    audio    2
AUDIOtrack pre-emphasis  copy-permitted tracktype channels
      4- 7          yes              no    audio    2
Table of Contents: total tracks:7, (total time 67:28.00)
  1.( 6:53.45),  2.( 7:08.30),  3.( 7:10.00),  4.(12:38.25),  5.( 4:43.50),
  6.(22:46.00),  7.( 6:08.00)
Is troll-adiposity coming from feederism?
With 24bit music you can listen to silence much louder!

DC sample for LocrianGroove tests

Reply #3 –
May it be possible one of these Vintage studio equipment had pre-emphasis implemented as spec so it is already in the signal and doesn´t change the noisefloor anymore?

In the song description they tell the hardware. Oberheim Matrix 12 S.R: Oberheim System with digital drums and Xpander, EMU 1, ARP and Casio
Is troll-adiposity coming from feederism?
With 24bit music you can listen to silence much louder!

DC sample for LocrianGroove tests

Reply #4 –
Quote from: 2Bdecided on
If the signal on the CD was encoded with pre-emphasis, you'd expect the noise floor to raise at higher frequencies, like this...
http://www.audioxpress.com/magsdirx/ax/add...ia/galo3025.pdf

Thanks for this link, David! I never understood what the point of pre-emphasis was, given the sufficient dynamic range of the CD spec.

Actually, I find the noise floor quite high at the high frequencies.

Comparing with some other CD masterings from that era, I'd say it sounds unusually bright, which makes me believe it was pre-emphasized during recording. Maybe it was pre-emphasized with analog equipment, i.e. before the A/D converter, maybe due to the reason mentioned in the PDF (bad ADC, hence also the offset?). That's my theory

Chris
If I don't reply to your reply, it means I agree with you.

DC sample for LocrianGroove tests

Reply #5 –
If I read correctly, Wombat's original intention was to de-emphasize his 1987 CD. So having read the link in David's post, I did a small experiment before going to bed  I used the transfer function data in the linked PDF to construct a 3rd-order IIR de-emphasis filter. That filter matches the PDF data to within 0.08 dB. I then applied the filter to the file posted above. The result is attached.

Wombat, can you post the result you got with sox so I can compare?

Chris

[attachment=5499:Western_..._De_Emph.flac]
If I don't reply to your reply, it means I agree with you.

DC sample for LocrianGroove tests

Reply #6 –
Quote from: C.R.Helmrich on
If I read correctly, Wombat's original intention was to de-emphasize his 1987 CD. So having read the link in David's post, I did a small experiment before going to bed  I used the transfer function data in the linked PDF to construct a 3rd-order IIR de-emphasis filter. That filter matches the PDF data to within 0.08 dB. I then applied the filter to the file posted above. The result is attached.

Wombat, can you post the result you got with sox so I can compare?

Chris

[attachment=5499:Western_..._De_Emph.flac]

Nice it found your interest.

Don´t know exactly what you want to archive. I normalised, removed DC and added dither so you can´t compare something.
If you only apply de-emphasis just use "sox --single-threaded input.wav output.wav deemph"

Edit 2: Grrrrr, 14.3.0 started to dither automaticaly. So if you only want to deemph you have to add -D
sox input.wav output.wav -D deemph

If you want to process it as i did just use:

sox --single-threaded "input.wav" "output.wav" --norm gain -1 highpass -1 5 deemph dither -f low-shibata stats

--single-threaded is used cause sox has a bug in 14.3.0 that may add clicks when applying deemph. --norm normalises the output. gain -1 decreases the input by 1dB to prevent possible clipping when applying the highpass to remove all DC. It is a 6dB filter at 5 Hz. i use a lower noise shaped dither cause i once played with amplified dither noise and liked this best if i remember right. I also trust blindly in Mr. Shibata
stats gives some info after processing.
If you have probs i may upload a version you want.

You may also compare your filter to sox and waveemph with the pcm_pre sample how i did in this thread. The Audacity graph shows.
http://www.hydrogenaudio.org/forums/index....25&start=25

Cheers!

Edit: One more thing. --norm doesn´t work on my windows machine with sox 14.2.0 so be sure to use 14.3.0
Is troll-adiposity coming from feederism?
With 24bit music you can listen to silence much louder!

DC sample for LocrianGroove tests

Reply #7 –
Quote from: C.R.Helmrich on
Comparing with some other CD masterings from that era, I'd say it sounds unusually bright...
I'd agree with that.

I think I was wrong about the noise floor. It is basically flat - but that region is still basically flat after it's been de-emphasised! I guess what looks like a broad region on a linear frequency view actually isn't broad enough to judge the slope of the spectrum. You'd need to see the noise down to about 3kHz to see the slope clearly. There's too much audio content in this range in this sample to judge.

Cheers,
David.

DC sample for LocrianGroove tests

Reply #8 –
Today i took me some time and did some relaxed listening on my main system. The de-emphasised version definitely sounds right and better. Where sometimes strange squichy squachy noise was it now sounds like it fits some instrument. Before it got confusing sometimes and sounded like electronic stuff. Now it sounds more like electronic instruments play on a virtual soundstage. Interesting how these sound-art New Agers arranged their stuff with pretty simple stuff compared to today. Sometimes underestimated or prejudged wrong. I really enjoy it. Most likely cause i spent so much time with it
Is troll-adiposity coming from feederism?
With 24bit music you can listen to silence much louder!

DC sample for LocrianGroove tests

Reply #9 –
Wombat, thanks for the sox command lines! Got it to work now. For the record, the maximum spectral magnitude difference between my upload and the sox result (sox -D in.wav out.wav deemph) is around 0.1 dB. Good enough for me, so you don't have to upload anything else

Btw, for me version 14.3.0 worked without using --single-threaded (or rather it made no difference). Maybe because I used a single-core Athlon, i.e. no Hyperthreading and stuff?

Chris
If I don't reply to your reply, it means I agree with you.

DC sample for LocrianGroove tests

Reply #10 –
Quote from: C.R.Helmrich on
I did a small experiment before going to bed  I used the transfer function data in the linked PDF to construct a 3rd-order IIR de-emphasis filter.

Wow! Did you pull tons of sliders in an equalizer or really did some mathematical thing and found a way to implement it? Funnily you just answered in a thread someone asked to do his own VST plugin. So a deemph VST plugin with input gain for easy use could be done. Of cause not by a noob like me...

Quote from: C.R.Helmrich on
Btw, for me version 14.3.0 worked without using --single-threaded (or rather it made no difference). Maybe because I used a single-core Athlon, i.e. no Hyperthreading and stuff?

Yes, has to do with multithreading as i have read on the sox bugtracker. They are aware of this prob and will fix that for sure.
Is troll-adiposity coming from feederism?
With 24bit music you can listen to silence much louder!

DC sample for LocrianGroove tests

Reply #11 –
Quote from: Wombat on
Wow! Did you pull tons of sliders in an equalizer or really did some mathematical thing and found a way to implement it? Funnily you just answered in a thread someone asked to do his own VST plugin. So a deemph VST plugin with input gain for easy use could be done. Of cause not by a noob like me...

I used Octave to calculate and apply the filter. No slider magic. But I recently found the Cool Edit 2000 effects plug-in API documentation, so if there is desperate need for a de-emphasis option in Audition 1.x, I am willing to write that plug-in. Would be a nice first step into writing such plug-ins, anyway, because the filtering process can be done in 10 lines of code.

Chris
If I don't reply to your reply, it means I agree with you.