Recommended Encoder Versions and Settings Options

[music_man_mpc]

I was just wondering why there is no one recommended setting for Vorbis anymore and where these advanced options came from? Are these advanced options an atempt from Xiph.org to merge third party tunings? Are there options that can resonably approximate the version of aoTuV used in the last multiformat test? And/or settings that can do the same for the last version of Megamix?

Thanks
-Tyler

[dawids]

I just subscribed to Jetaudio 6 plus to enter the FLAC world (highest processing -8), and own MMJB as well (previously used for MP3pro VBR 100% ripping from CD). Winamp does not allow registration in SA for some reason. Are all these switches and settings applicable with these applications, or will I be getting things like microattack?
Comments / guidance appreciated.

[QuantumKnot]

Thanks for all your help, QK!  It really helps us newbies out!

One more question:
Regarding the 'impulse_trigger_profile' setting, you said:

Note 1: There is the possibility that relaxed block switching may cause other quality problems and create suboptimal Vorbis files.  So use sparingly and with caution.

I'm not sure what you mean by 'relaxed block switching'. Are you referring to a lower profile number in the 'impulse trigger profile' string (ie, --advanced-encode-option impulse_trigger_profile=0 (or 1))? In other words, are you recommending the use of higher ITP values?

The higher the profile, the more 'relaxed' block switching is.

I was intending to use an ITP value of 3 for most (if not all) of my files, but your warning to 'use sparingly and with caution' caused me to balk.  Any clarification would be great!

Ah ok. I recently heard reports that using a higher ITP can cause quality problems in some cases.

So ITP should be used only if there is a noise problem and you want to try and fix it. But you shouldn't use or expect ITP to improve the quality of your music in general. It is a noise problem fixer, not a general-purpose quality improver In 95% of the cases, ITP on default is recommended. I need to rethink whether I should even mention ITP in a thread about recommendations.

INT (impulse_noisetune) however seems safe to use in general and I haven't heard reports of INT causing quality problems (yet).

Also, I noticed that in the 'Vorbis History' section, it still reads:

...John33 merged the sources to give us GT3b2, which is now the recommended Vorbis encoder.

Does this still hold true in light of the release of 1.1?

Looks like I need to do some more editing. Thanks

This post has been edited by QuantumKnot: Sep 29 2004, 00:38

[esa372]

Thanks again, QK, for all the info..! It's really helping me dial in my settings...



~esa

This post has been edited by esa372: Sep 30 2004, 20:34

[Prodoc]

@QuantumKnot:

INT (impulse_noisetune) however seems safe to use in general and I haven't heard reports of INT causing quality problems (yet).

From main post:

Therefore, you should try a small value to start off (say -5) and see if you get acceptable quality.

So does this mean that the recommended setting for INT is -5?
You might want to specify this a bit incase everyone starts using -5 while they shouldn't
I, for one, wouldn't know which value I should use though I do think it's important to set this option. What would be the best value to use in general for all encodings?

[QuantumKnot]

@QuantumKnot:

INT (impulse_noisetune) however seems safe to use in general and I haven't heard reports of INT causing quality problems (yet).

From main post:

Therefore, you should try a small value to start off (say -5) and see if you get acceptable quality.

So does this mean that the recommended setting for INT is -5?
You might want to specify this a bit incase everyone starts using -5 while they shouldn't
I, for one, wouldn't know which value I should use though I do think it's important to set this option. What would be the best value to use in general for all encodings?

It depends really. Usually you don't have to use INT in most cases. But if you are sensitive to pre-echo and can hear it in some music, INT will be useful here. But if you can't hear pre-echo at default settings (like me, I rarely am able to hear pre-echo unless it is really sharp stuff like castanets or harpsichord), then there is no need to set INT to anything. I usually set it to -5 for safe measure, even though I won't be able to tell the difference most of the time.

I suggest you take a sample that is representative of the music you listen to most, do some blind listening tests to see if pre-echo is an issue....then gradually increase INT (or more like decrease since its negative) until you feel the pre-echo is gone. Then if the bitrate jumps are acceptable, then you can use that.

This post has been edited by QuantumKnot: Oct 6 2004, 02:59

[LoFiYo]

If this can be considered off-topic, I apologize in advance.

I was just reading some comments by Aoyumi on 1.1 at his website.

He regrets that he wasn't taking a close look at the official Vorbis as his code was being incorporated into it by Monty, and expressed a slight degree of concern regarding the quality of 1.1 as compared to AoTuV b2.

He suspects that it is technically possible that some parts of AoTuV b2 tuning may have been broken when Monty fixed the code of AoTuV b2 and incorporated it into the official Vorbis, because Aoyumi, when tuning AoTuV, had also been taking into account whatever Monty fixed.

So despite many people's assumption, the audio quality of 1.1 is not automatically the same as that of AoTuV b2.

[Madman2003]

@QK: higher ITP are normally used with higher quality levels, why would they cause problems? (i can see that lowering could be bad, but why is raising them?)

[QuantumKnot]

@QK: higher ITP are normally used with higher quality levels, why would they cause problems? (i can see that lowering could be bad, but why is raising them?)

Nobody knows, at present. There was one report of it hurting quality but that's the only one. It's not enough to be conclusive but it's better to be safe for now until it is proven/disproven.

Besides, I made ITP to fix microattack problems rather than to improve quality in general use.

[Enig123]

As we know ITP uses short blocks more frequently, maby this suffers the frequency resolution?

[Gray_Wolf]

Hi, i am very happy with the sound quality of this new version of vorbis 1.1.0 encoder; I tested with many samples for my music archive quality backups. My favorite setting is -q7 (around 224kbps average) the sound is very clean and good taste for my ears ; and not differences with the original wav file... I tested the setting -q0 for radio stream use, and the quality is good too. I am using the oddcast plugin dsp_oddcast_02262003; the question is: How i can change the old vorbis encoder of this oddcast plugin for the new version 1.1.0?

Other problem for me is the new version of oggdropXPd (1.7.11-1.1.0) ; i am running windows 95; and my system completely crash with this version of oggdropXPd.
I am using Oggifier frontend with the oggenc.exe (1.1.0) without problems , but; Is this oggenc.exe the same encoder in the oggdropXPd combination?

I am very appreciated some feedback and help

[QuantumKnot]

Hi, i am very happy with the sound quality of this new version of vorbis 1.1.0 encoder; I tested with many samples for my music archive quality backups. My favorite setting is -q7 (around 224kbps average) the sound is very clean and good taste for my ears ; and not differences with the original wav file... I tested the setting -q0 for radio stream use, and the quality is good too.  I am using the oddcast plugin dsp_oddcast_02262003; the question is: How i can change the old vorbis encoder of this oddcast plugin for the new version 1.1.0? 

Other problem for me is the new version of oggdropXPd (1.7.11-1.1.0)  ; i am running windows 95; and my system completely crash with this version of oggdropXPd.
I am using Oggifier frontend with the oggenc.exe (1.1.0) without problems , but; Is this oggenc.exe the same encoder in the oggdropXPd combination?

I am very appreciated some feedback and help

Yes, oggenc is the same encoder as oggdropXPd. That is, they produce the same version of Vorbis files.

[Gray_Wolf]

Yes, oggenc is the same encoder as oggdropXPd.  That is, they produce the same version of Vorbis files.

Thanks QK

[Patrick00]

vorbis 1.1.0 -q7 great quality, transparent to my ears. just for kicks i tried q0, great quality for its bitrate. (used placebo - days before you came) - largest distortion was the muddled guitars - lol, im glad i use q7

[Gray_Wolf]

vorbis 1.1.0 -q7 great quality, transparent to my ears. just for kicks i tried q0, great quality for its bitrate. (used placebo - days before you came) - largest distortion was the muddled guitars - lol, im glad i use q7

I mentioned the setting -q0 only for radio internet broadcast; not for high fidelity.

The -q0 switch is in average 64 kbps, and the sound quality is better than 128kbps mp3 radio internet broadcast for the half of the bandwidth .

It's the ideal solution for internet radio broadcasters, and not paid for license, because vorbis is patent free.

This post has been edited by Gray_Wolf: Oct 25 2004, 12:23

[Poromenos]

Actually -q-1 sounds better than mp3 128 to me... I wonder why people don't use it for streaming, and why you can't do custom bitrates, like 24/32/whatever. Are there technical limitations?

[Patrick00]

I mentioned the setting -q0 only for radio internet broadcast; not for high fidelity.

The -q0 switch is in average 64 kbps, and the sound quality is better than 128kbps mp3 radio internet broadcast for the half of the bandwidth .

It's the ideal solution for internet radio broadcasters, and not paid for license, because vorbis is patent free.

i realize this, i was just testing it out for the sake of testing, i'm not THAT low on space . i was just commenting on the suprising quality, as the previous user said, "q-1 sounds better than mp3 128" - in some cases.

[Gray_Wolf]

i realize this, i was just testing it out for the sake of testing, i'm not THAT low on space  . i was just commenting on the suprising quality, as the previous user said, "q-1 sounds better than mp3 128" - in some cases.

OK

[[solid]]

erm... don't shout on me please...
i seriously doubt if i could abx even q4, but can one emulate gt3b2 with these advanced options? what should they be set to get something similar to gt3b2 at q6?

[QuantumKnot]

erm... don't shout on me please...
i seriously doubt if i could abx even q4, but can one emulate gt3b2 with these advanced options? what should they be set to get something similar to gt3b2 at q6?

It is rather difficult at q 4 since GT3b2 was only tuned for q 5 and above.

[[solid]]

no no, i've meant:
gt3b2 at quality 6 - can a combination of the advanced options present in the recommended encore reproduce that tuning?
edit: the allnew rewritten from scratch version 2.0 of my post

This post has been edited by [solid]: Nov 15 2004, 22:54

[Dologan]

no no, i've meant:
gt3b2 at quality 6 - can a combination of the advanced options present in the recommended encore reproduce that tuning?
edit: the allnew rewritten from scratch version 2.0 of my post 

I'd say oggenc -q 6 --advanced-encode-option impulse_trigger_profile=5 would be equivalent, although not bit identical, since GT3b2 does not include aotuvb2's tunings, as 1.1 does.

[QuantumKnot]

no no, i've meant:
gt3b2 at quality 6 - can a combination of the advanced options present in the recommended encore reproduce that tuning?
edit: the allnew rewritten from scratch version 2.0 of my post 

Sorry, I misread it.

I've compared the critical values between 1.1 and GT3b2 and I think these settings should be close enough for q 6:

CODE

--advanced-encode-option impulse_noisetune=-10 --advanced-encode-option impulse_trigger_profile=5.2

[Zoom]

I just built an AMD64 system and was updating my software. I tried both of the ICL compiles (P3/AMD & P4) and found that the P4 version was consistently around 2x faster. (*Not double speed)

CODE

P4 ICL Compile
Done encoding file "M:\My Music\Track03.ogg"

       File length:  2m 39.0s
       Elapsed time: 0m 08.0s
       Rate:         19.9717
       Average bitrate: 167.0 kb/s

P3/AMD
Done encoding file "M:\My Music\Track03.ogg"

       File length:  2m 39.0s
       Elapsed time: 0m 09.0s
       Rate:         17.7526
       Average bitrate: 167.0 kb/s

So should you use the P4 compile if you have AMD64 or should you still use the P3 version?

[kotrtim]

......the bitrate is similar, 167kbps, so they are bit identical.......why bother, go for the faster compile

i think the P4 means SSE2 optimised, P3 means SSE optimised, nothing important

If you can run the P4 compile, that means AMD64 supports SSE2