VideoHelp Forum
+ Reply to Thread
Results 1 to 6 of 6
Thread
  1. Hi,

    I have got the same song twice, but the one is ripped in surround sound and the other in stereo.

    It looks like this in Spek:

    Click image for larger version

Name:	FLAC.PNG
Views:	719
Size:	1.19 MB
ID:	39933

    The 5.1 file shows much lower statistics, but the 5.1 is about 3100 kBit/s and the 2.0 about 800. I'm not familiar with audio files, so can you tell me why this is?

    Btw. both are level 0 compressed.
    Quote Quote  
  2. Dinosaur Supervisor KarMa's Avatar
    Join Date
    Jul 2015
    Location
    US
    Search Comp PM
    Originally Posted by draig-llofrudd View Post
    The 5.1 file shows much lower statistics
    Don't know what you mean by this but if you are talking about the high end of the spectrum, I suspect the signal is still there but you just have to drop the cutoff in Spek. Ctrl + DownKey to drop it to 140dB.
    Last edited by KarMa; 15th Dec 2016 at 18:20.
    Quote Quote  
  3. Seem 5.1 is less louder and as 24 bit free from noiseshaping. You should align (level) on both - 120dB is approx 20 bit dynamics.
    Quote Quote  
  4. Yes, I mean the high end of the spectrum. I tried a bit around with Ctrl + Up/Down, but the 5.1 is significantly lower in high-end. I couldn't notice a noise difference. I guess some subliminal sounds are different here. Ok, not that important. But what do you mean by noiseshaping? What is that sky-ish looking area in top of the right graphics? The song was ripped from the same CD, one time 5.1 and the other time 2.0 - and it originally is 5.1.

    Btw you say 120 dB is good for 20 bit. What shall I use for 16 and what for 24?
    Quote Quote  
  5. Originally Posted by draig-llofrudd View Post
    Yes, I mean the high end of the spectrum. I tried a bit around with Ctrl + Up/Down, but the 5.1 is significantly lower in high-end. I couldn't notice a noise difference. I guess some subliminal sounds are different here. Ok, not that important. But what do you mean by noiseshaping? What is that sky-ish looking area in top of the right graphics? The song was ripped from the same CD, one time 5.1 and the other time 2.0 - and it originally is 5.1.

    Btw you say 120 dB is good for 20 bit. What shall I use for 16 and what for 24?
    Well i can confirm your concerns about high frequency energy - those small vertical stripes seem to be higher on CD signal version but i have impression that overall signal level is also higher - you can try to check amplitude statistic - difference between two signal versions (peak to peak and RMS) should be quite significant. I would align both signals before spectral analysis.

    Noise shaping is a way to reduce quantization noise at desired part of audible spectrum at a cost of increase of overall noise - this is exactly
    Originally Posted by draig-llofrudd View Post
    sky-ish looking area in top
    - as CD use 16 bit and source usually has higher bitdepth then signal need to be re-quantized - to reduce error introduced in a requantization process, additional noise is added (so called dither) and noise shaping is used to improve (reduce perceived noise) quality.

    Amplitude statistic should explain differences (i think CD may have reduced dynamics by gain compression and overall signal levels higher - clearly different spectrum).

    Bellow is SoX https://sourceforge.net/projects/sox/ spectrum analyzer script - we should be able to see whats going on bellow -120dBFS.
    Code:
    @set dyna=180
    @set /a clut=%dyna%/5
    
    @sox --multi-threaded --buffer 524288 -S -V -D "%1" -n spectrogram -z %dyna% -w Kaiser -q %clut% -y 1025 -x 2048 -s -o "%~n1_%dyna%dB.png" stat stats -b 16
    Last edited by pandy; 16th Dec 2016 at 14:29.
    Quote Quote  
  6. The program used to export the stereo flac file used dithering -- hence the high frequency noise.
    Quote Quote  



Similar Threads

Visit our sponsor! Try DVDFab and backup Blu-rays!