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: Ogg killer track (Read 11985 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Ogg killer track

Reply #25 –
Monty should know about this, if he doesn't already know.  Especially given such clear ABX results at -q 9.

See above for clip download info, ff123's site.
God kills a kitten every time you encode with CBR 320

Ogg killer track

Reply #26 –
I can't hear anything wrong at -q6 :-/

--
GCP

Ogg killer track

Reply #27 –
Quote
Originally posted by timcupery
Monty should know about this, if he doesn't already know.


xiphmont in the thread above = Monty

--
GCP

Ogg killer track

Reply #28 –
Ok, I tried to simulate the distortion effect, so that more people (especially Vorbis devs) could hear this. Unfortunately going to lower quality settings didn't make this effect much louder, so I had to play with CoolEdit...

My simulated distortion goes "out of rhytm" around 7secs (which doesn't quite sound like the actual encoding), so it should be easiliy noticeable (I still didn't make it very loud). 0-5 sec sounds pretty much like actual encoding but with increased distortion.

This does not sound exactly like the actual distortion (the actual is somewhat lower frequency, more "muddy" and more silent) but it's hopefully close enough to give some hints what I (and probably gambito) hear. Of course the actual distortions are clearly more silent, but still very noticeable after you start hearing it.

http://sivut.koti.soon.fi/julaak/Queen_OggDistortion.flac

Btw. Garf's RC2GT2 160kbps produces clearly less this type of distortion than RC3 -q 6.
Juha Laaksonheimo

Ogg killer track

Reply #29 –
I tested some of the basic MPEG committee test tracks and Vorbis has problems with mouth-organ sample.

Here's part of it.
http://sivut.koti.soon.fi/julaak/mouth_organ.flac

-q 5 especially could be better in my opinion. -q 6 I ABXed quickly 15/16. Problem here is quite rough sound.
Juha Laaksonheimo

Ogg killer track

Reply #30 –
Well, this is a few days old clip, but just to collect some problem clips together..

Daniel Resare found this clip, the distortion is a bit similar with Queen clip (maybe both can be fixed at the same time).
"In the beginning of the provided sample the artifact is easily noticed in the left channel, but later in the sample i belive it is somewhat to the right in the stereo image.

The artifact dissapears somewhere around quality 9.0 and it does not
exist in rc2."

http://noa.tm/vorbis/left-dist/

The distortion is so obvious especially in the beginning that there's no need to ABX -q 6 quality level.
Juha Laaksonheimo

Ogg killer track

Reply #31 –
JohnV, thanks for your work on this.
God kills a kitten every time you encode with CBR 320

Ogg killer track

Reply #33 –
Yeah, Garf and I tried to get Monty to hear a clear artifact in it: one clear artifact is when Vega sings "says *the* man...".
*the* is distorted and spreads the noise.

[edit]
Speek's tomsdiner.wav hosted here as FLAC-format:
http://static.hydrogenaudio.org/extra/samp.../tomsdiner.flac
Juha Laaksonheimo

Ogg killer track

Reply #34 –
Ok, I'm adding ogpulse.wav found by MrDrew, to this list.
The problem is that higher tone (toot toot) sounds after 2 second are quite rough and can be relatively easily ABXed even with -q 10.

MrDrew's original thread:
http://www.hydrogenaudio.org/forums/showth...=&threadid=1111
Juha Laaksonheimo

Ogg killer track

Reply #35 –
Okay... I posted the FLAC version of the OGPULSE sample at the following site:

http://personal.www.umich.edu/~aklenk/music/ogpulse.flac

One thing that should be known about this sample (given that it may be used to tune Vorbis) is that was not ripped off of a CD.  It came from an MP3 that I had.  Granted, it's a 320kbps mp3, and it sounds (and looks in Cool Edit) completely artifact free.  Still I thought people might want to know that    Here's the info on that mp3:


Milk Inc - Walk On Water.mp3
-------------------------------

Bitrates:
----------------------------------------------------
320    ||||||||||||||||||||||||||||||||||||||||      100.0%
----------------------------------------------------

Type                : mpeg 1 layer III
Bitrate            : 320
Mode                : stereo
Frequency          : 44100 Hz
Frames              : 7277
Length              : 00:03:10
Av. Reservoir      : 0
Emphasis            : none
Scalefac            : 7.2%
Complete            : yes
Encoder            : FhG (ACM or producer pro)

--[ EncSpot 1.1 ]--[ http://www.guerillasoft.com ]--



-- Andrew
 

Ogg killer track

Reply #36 –
Ok, here's a clear lossy channel coupling problem found by Wizard.

Problem with lossy channel coupling (-q 4.99) is that the flanging on the left channel is clearly reduced or moved to the middle. This is especially clear near 1.6sec and 3.0sec where the flange is moved towards middle. -q 5 's lossless channel coupling sounds correct.
I verified this with several blind listening tests.

Sample:
http://sivut.koti.soon.fi/julaak/flange_499vs500.flac

Original thread: http://www.hydrogenaudio.org/forums/showth...=&threadid=1255
Juha Laaksonheimo