| Author |
 Topic 
|
PerStromgren
Location: Sweden
Posts: 8
|
 Posted - Thu Apr 03, 2003 1:50 am
|
|
|
I use my CoolEdit 2000 for anlysing music dynamics, and got noise figures around -65db from the statistics tool. To calibrate things I ripped a test CD track that has digital silence (i.e. only dither noise) and asked Cooledit to analyze the track. It says min and max RMS value around -70dB! I would have though I would get somewhere around -93dB or so. I also tried a sinus signal with a peak value of 0dB FS, -10db, -20 dB, etc down to -90, and when the signal goes under -60dB I get a lot of noise. Where does this noise come from?
I suppose that when I record a CD into Cooledit, no analog stages are used, is that right?
Per.
|
|
AndyH
Posts: 1425
|
|
Posted - Thu Apr 03, 2003 2:30 am
|
|
|
dither noise is not silence, it is indeed noise.
If you generated tone signals in CE at various decreasing levels, as you indicated, and if you began hearing noise that was not part of your generated tones as the signal level of those tones went lower, the noise came from your playback system. That means your soundcard and whatever other electronics you used to turn the CE tone files into sound.
|
|
PerStromgren
Location: Sweden
Posts: 8
|
|
Posted - Thu Apr 03, 2003 2:44 am
|
|
|
| AndyH wrote: |
| dither noise is not silence, it is indeed noise. |
Yes, sure, but 25dB dithering noise, are they mad? That would really be a waste of headroom, wouldn't it? I'm not saying that I hear the noise, I'm saying that CE tells me these levels.
I generated silence with CE, and this signal is indentified correctly as silence from the statistics tool.
Ref: HiFi NEWS & Record Review Test Disc III, track 74.
Per.
|
|
Andrew Rose
Location: United Kingdom
Posts: 875
|
|
Posted - Thu Apr 03, 2003 2:56 am
|
|
|
|
Have you done a frequency analysis to see where the noise is. Chances are it's in the upper frequencies where it's less likely to be heard. The effective result of this is a much lower perceived noise floor, which is precisely how dithering works. By shifting the noise away from the frequencies to which we're most sensitive to those we don't hear so well the effect is to make the noise floor appear lower. Crude measurements of this don't tell you as much as your ears might...
|
|
post78
Location: USA
Posts: 2887
|
|
PerStromgren
Location: Sweden
Posts: 8
|
|
Posted - Thu Apr 03, 2003 3:59 am
|
|
|
| Andrew Rose wrote: |
| Have you done a frequency analysis to see where the noise is. Chances are it's in the upper frequencies where it's less likely to be heard. |
Good idea! I just did, and it appears that there is "bass tilt" with the signal being lifted 30 dB at the left hand end of the scale (10 Hz or so). The clip is 10 seconds, should be enough to avoid border effects.
No, it does not seem to be dithering, what else could it be, a faulty CD ROM unit, perhaps? It works flawlessly when reading data, does it work in a different way when reading red book CDs?
Per.
|
|
post78
Location: USA
Posts: 2887
|
|
Posted - Thu Apr 03, 2003 4:16 am
|
|
|
Do you have a remote website that you can upload a picture to? If so, do the following:
Open the frequency analyzer in CE and double-click the waveform to select the entire file (or the other way around, it doesn't matter). Next, click anywhere on the analyzer to be sure that particular window is selected, then on your keyboard hit "Alt - Print Screen". Next, open a simple image editor and select "copy > paste". Save the file and upload it to your website, then give us a link to view it.
If you don't have a remote site, than you can e-mail it to me and I will upload the image. I won't be able to do this until tomorrow though, so you might already have an answer by then.
_________________
Answer = 1. Probably.
|
|
|
|
post78
Location: USA
Posts: 2887
|
|
Posted - Thu Apr 03, 2003 4:18 am
|
|
|
Oh yeah, use a fairly large FFT size.
_________________
Answer = 1. Probably.
|
|
|
|
ozpeter
Location: Australia
Posts: 3200
|
|
Posted - Thu Apr 03, 2003 5:11 am
|
|
|
| Quote: |
| I suppose that when I record a CD into Cooledit, no analog stages are used, is that right? |
Depends very much on how you 'record' it - you've used the word 'rip' earlier in your post, which would imply a digital transfer, but the word 'record' could involve your soundcard and possibly analog stages, depending on your system's configuration. So for the sake of clarification, exactly how did you transfer the CD track to your PC?
- Ozpeter
|
|
PerStromgren
Location: Sweden
Posts: 8
|
|
Posted - Thu Apr 03, 2003 6:46 am
|
|
|
| Quote: |
| Do you have a remote website that you can upload a picture to? |
Yes, here it is: http://w1.541.telia.com/~u54104547/cooledit.htm.
I don't know what you consider a "fairly large" FFT size, I choose 8192, I hope that is enough.
Per.
|
|
PerStromgren
Location: Sweden
Posts: 8
|
|
Posted - Thu Apr 03, 2003 6:52 am
|
|
|
| Quote: |
| So for the sake of clarification, exactly how did you transfer the CD track to your PC? |
This is what I did:
1. Open Cool Edit (CE)
2. Insert CD into computer
3. Click the red "Record" button in the CE window
4. Select 44kHz/16bits Stereo and click OK
5. Click a track in the CD player section to play CD
6. Hit stop in both sections after 20 seconds
This most be a digital transfer, must it not?
Per.
|
|
Gulliver
Location: Estonia
Posts: 442
|
|
Posted - Thu Apr 03, 2003 7:28 am
|
|
|
In this case you just recorded your soundcard's noise floor, I'm afraid... 
|
|
Andrew Rose
Location: United Kingdom
Posts: 875
|
|
Posted - Thu Apr 03, 2003 7:33 am
|
|
|
| PerStromgren wrote: |
| This most be a digital transfer, must it not? |
No - it sounds like it's coming out of your CD-ROM as an analogue file and being redigitised by your soundcard - hence you're relying on your CD-ROM's DACs and your soundcards ADCs.
If you get a hold of something like Exact Audio Copy and rip the CD into your PC then you'll have a digital copy.
|
|
Bobbsy
Location: United Kingdom
Posts: 327
|
|
Posted - Thu Apr 03, 2003 7:39 am
|
|
|
Nope, using that method there's no guarantee at all that you have a digital copy...in fact it's most likely not. Chances are you're going through a D to A then A to D chain to get the material into CE.
You don't mention either what brand of sound card or what Windows Mixer settings you've used. Some sound cards give you a choice of "CD Audio" and "CD Digital" on the record mixer, others don't give you this choice and only do analogue audio.
However, by far the best way to transfer CDs into CE is to use an actual ripper (and no, what you did was NOT ripping...you were just doing a basic recording).
There are lots of rippers out there, but one of the best happens to be freeware. Go to http://www.exactaudiocopy.de/ and download Exact Audio Copy. When you use this to transfer CD data onto your hard drive it stays in pure data form and is not processed at all by your sound card. Besides quality, it also does this faster than real time, so you have a win/win situation!
Anyhow, give EAC a try and repeat your tests. I suspect strongly you'll get very different results when you rip your CD rather than recording.
Cheers,
Bob
|
|
|
|
Andrew Rose
Location: United Kingdom
Posts: 875
|
|
Posted - Thu Apr 03, 2003 8:04 am
|
|
|
Snap! 
|
|
Syntrillium M.D.
Location: USA
Posts: 5124
|
|
Posted - Thu Apr 03, 2003 8:23 am
|
|
|
Hello Per. As you've probably summized (from the previous posts), you did not do a direct-digital transfer, but rather a Digital-to-Analog transfer, which in turn, records the noise of your soundcard.
It would also appear (from your snapshot) that your card is exhibiting a little DC Offset (not uncommon with on-board cards and soundblasters) - this 'can' be corrected, but that might very well explain the 'ramp up' in the 1-10Hz range.
---Syntrillium, M.D.
_________________
|
|
|
|
| |
Topic
|
All Forums
Digital silence is only -70dB!
