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: Noise tonality (Read 4735 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Noise tonality

Is it a bad sign if an encoder can't reproduce the same tonality in a pure white noise sample like in the original?

I encoded for fun white noise samples and noticed while WMA9 kept the same subjective tonality, Ogg 1.0 has a lil' higher tonality (I would call it pitch right now, but I wont  ).

Is that of any importance?

Noise tonality

Reply #1 –
Eh.. If I'm not completely mistaken, pure white noise is 100% noisy, 0% tonal. So there's no "tonality" in that regard in pure white noise, or the tonality is complete noise..
Juha Laaksonheimo

Noise tonality

Reply #2 –
Uh, well I just don't find the right way to express myself. Well the Ogg noise sounds brighter than the original and WMA9 noise.

Noise tonality

Reply #3 –
I know what you're saying, it's like listening to white noise through different loudspeaker brands/models.  It doesn't sound the same through each one, because there's a different voice character to each speaker.  I don't think it's a test worth worrying about, since the codecs are designed toward keeping *music* sounding faithful and the methods they use to do that would quite possibly be thwarted with a white noise signal.

Noise tonality

Reply #4 –
I agree that how a lossy encoder performed on white noise would not be a definitive test, but I do think that for a encoder to be transparent on music it would have to be transparent on white noise too.

Unlike some of the other torture samples mentioned from time to time, noise is actually an important component of real music. In fact, some instruments (like cymbals) generate almost exclusively noise (with the different sizes having different "tonalities").

Noise tonality

Reply #5 –
And not to forget historic performances!
Furthermore classical music usually has a well noticeable noise level (due to higher dynamics and difficulties in recording huge orchestras).

IMO exact noise encoding is essential for transparence!

Noise tonality

Reply #6 –
I would think most instruments, even ones that sound "noisy" are pretty far from pure, gaussian white noise..

If you want to see what's happening to your noise with the encoders you're testing..
find one of the spectrogram programs out there,  and take a look at the results.

Remember, something can look drastically different on a spectrogram (usually at the high end) and sound very unchanged,  and vice versa..  (usually at the lower end)

Noise tonality

Reply #7 –
Quote
Originally posted by Tom Servo
Is it a bad sign if an encoder can't reproduce the same tonality in a pure white noise sample like in the original?

I encoded for fun white noise samples and noticed while WMA9 kept the same subjective tonality, Ogg 1.0 has a lil' higher tonality (I would call it pitch right now, but I wont  ).

Is that of any importance?


Perhaps, if you intend to encode noise for the rest of your live...

Noise tonality

Reply #8 –
Don't know about encoding white noise, but the way vorbis alters background noise in some samples affects overall quality of the resultant encode.

Noise tonality

Reply #9 –
Perhaps a bit of a leap here, but when I encoded a recording from an old tape and let it run thru lame (alt preset insane), all the hissing was gone.

Noise tonality

Reply #10 –
After doing the 64kbps listening test, I've concluded that vorbis emphatized the highs, so it might be what Tom is saying, and Indeed is of importance.
 

Noise tonality

Reply #11 –
Let's wait for the interim, or better yet, final results before jumping to any conclusions. The test is still open.

Noise tonality

Reply #12 –
After ff123's call for samples I assembled a bunch of 21 samples that imho represent my music taste pretty well without being codec killers. Unluckily I was too late, but I now use them for my own testing. Yesterday I tried out ogg -q 4 (quick and dirty, no abx or anything) which in all cases I could distinguish from the original due to it's emphasized highs. A sample with short "blips" (Cirillo - Cristallo) in it also had heavy preecho. So I think Jaz's conclusion is valid. But overall I was impressed by the quality considering the bitrate. If you want I can find a way to host a few of them (the whole bunch is about 55mb losslessly compressed).
Some examples:
Gouryella - Tenshi (original mix)
a-ha - Afternoon High [album: lifelines]
Leahy - B Minor
Electrasy - Foot Soldiers [album: in here we fall]

Noise tonality

Reply #13 –
I do not intend to encode white noise, especially coz it's not entertaining as music style. But if the noise pitch/brightness (or whatever I should call it) does differ slightly from the original, it could maybe also affect hihats and the like, or noisy samples in electronic music. Kinda same when I activate the PNS tool in Psytel, hihats and noise in general are also screwed.