Audacity bug?

[TBone]

the following plot captures level 5 (hardest) 100hz vertically & lateral cut tracking test ...

notice the +20 Y axis, below changed to 0dB ...

Numbers were exported from Audacity(v2.1.0), example below:

Audacity 180_5L track, shows no clipping ...

... however the freq.plot show +2.1dB@101hz ...

Little confused ... ?

Audacity 2.1.0 exports positive numbers ... ?
freq. dB
97.5723 -29
98.2452 -24
98.9182 -5
99.5911 3
100.2640 0
100.9369 -9
101.6098 -20
102.2827 -27

 

[RayDunzl]

Little confused ... ?

It seems to combine the tracks - and maybe be a little buggy on the side

1000Hz tone mono full scale reads +0.4db
1000Hz tone stereo full scale reads +6.4db

(you might also get slightly different numbers with different window types)

Or maybe it is right... Look up 'intersample over'.

 

[TBone]

thx Ray ...

found some correspondence concerning that subject @audacity.

"As we are all aware, audio signals can be thought of in terms of sine
> wave components and phase (FFT analysis). The maximum possible size of
> the inter-sample peak for any sine tone is related to the frequency.
> Only with very high frequencies do we see large inter-sample peaks -
> for low and mid-range frequencies the inter-sample peaks are always
> extremely small."

at least that makes some sense, the graph shown above is a difficult test and should include some tracking distortion/artifacts.

ps. I can separate the tracks, and probably should, but I was stumped by the initial positive results.

 

[amirm]

FFT values depend on accuracy of the math and windowing used. I am not sure how good those are in Audacity. Do you want to share the file with me and I use Audition on it?

 

[TBone]

another track, left right, all negatives ...

-36.8526 -36.420128
-24.63866 -25.31794
-9.232555 -9.712951
-3.963328 -4.457605
-4.344807 -4.835027
-9.620387 -10.082705
-24.981867 -25.6054
-35.624516 -34.802303
-38.186714 -37.939274

mono, positives ...
-33.030396
-31.3526
-27.694687
-18.913363
-3.108192
2.22338
1.716651
-3.667533
-18.551216
-27.28858

 

[Blumlein 88]

Yes for one thing you are getting the two channels combined. Simply split them to mono and you will find the FFT on one channel matches the levels for the most part.

 

[amirm]

Here the same status from Adobe Audition:

 

[amirm]

And the same stats, but this time with Audition converting the file to mono/one-channel:

 

[RayDunzl]

Wow...

Dynamic Range around 340dB...

 

[TBone]

Ray, reminds me of the Ramones live; constant 100hz tones minus 340dB.

thanks for your time, Amir.

 

[amirm]

Yeh, seems like some silliness with minus infinity for minimum RMS power. Likely that error led to the 340 db dynamic range value.

 

[Blumlein 88]

-340 db is only 57 bits more or less. Any good 64 bit DAC should handle it fine. We have some 32 bit DACs so maybe you need two of those for 64 bit. Preferably NOS multi-bit versions.

 

[RayDunzl]

Anyone working on a Qubit DAC?

 

[Blumlein 88]

Anyone working on a Qubit DAC?

Should be the ultimate for DSD.