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: psymodel 1 and 2 of toolame (.mp2) (Read 8228 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

psymodel 1 and 2 of toolame (.mp2)

Hey 

Does anyone know the difference between psychoacoustic model 1 and psychoacoustic model 2 of toolame?

I want to encode a movie (DVD to VCD) and I am not sure which model is better to use... I believe one of the models is for music and the other is for speech, though I do not know which is which....

Also, also which model would be the better for movies: the music model or the speech model?

Thanks for help

psymodel 1 and 2 of toolame (.mp2)

Reply #1
Quote
Does anyone know the difference between psychoacoustic model 1 and psychoacoustic model 2 of toolame?
All I know is this: model 2 crashes my Win ME box.

psymodel 1 and 2 of toolame (.mp2)

Reply #2
Psymodel 1 and 2 differ in complexity as well as in way of computing of the masking threshold. Psymodel 2 is more complex (slower) than psymodel 1.

At 192 kbps I found psymodel 2 much better for the set of samples I used.

ISO recommends psymodel 1 for MPEG Layer 1 and 2 and psymodel 2 for Layer 3, but that could be wrong, of course.

psymodel 1 and 2 of toolame (.mp2)

Reply #3
Quote
Originally posted by Ivan Dimkovic
Psymodel 1 and 2 differ in complexity as well as in way of computing of the masking threshold. Psymodel 2 is more complex (slower) than psymodel 1.

At 192 kbps I found psymodel 2 much better for the set of samples I used. 

ISO recommends psymodel 1 for MPEG Layer 1 and 2 and psymodel 2 for Layer 3, but that could be wrong, of course.


In tooLame psymodel 2 works much better than psymodel 1. I recommend to use psymodel 2,
computational complexity is irrelevant for
todays computers.

But also psymodel 2 has a lot of flaws.

Basic problem of MPEG LAyer 1 and 2 are the
insufficient psychomodels used when they were introduced.
--  Frank Klemm

psymodel 1 and 2 of toolame (.mp2)

Reply #4
Quote
Originally posted by Frank Klemm
Basic problem of MPEG LAyer 1 and 2 are the
insufficient psychomodels used when they were introduced.

while we're at it - I wonder whether we will see a mppenc capable of writing mpeg2-streams anytime soon? :eek:

I am much more interested into that feature than into SV8, but that's just me..
A riddle is a short sword attached to the next 2000 years.

psymodel 1 and 2 of toolame (.mp2)

Reply #5
Thanks to everyone for their replies :-)

I emailed Mike Cheng about it and he might post here... or I shall post here if he responds...

In cool edit 96 the precursor to the might cool edit 2000, the mpeg 1 layer 2 option tab said that:

psymodel 1 = MUSICAM
psymodel 2= AT&T

What does that mean?

In toolame psymodel 1 is the default, so one would think it is better.  However, psymodel 2 does take twice as long to encode, so that seems to support Ivan and Frank's comments.

PS I second samadi's idea for SV8 of mpc to allow .mp2 output...

psymodel 1 and 2 of toolame (.mp2)

Reply #6
nah, that's not exactly what I intended to say 

Frank talked about a modified mppenc and a tool which creates mpeg2 audio streams from that mppenc's output a while ago (iirc), and I am wondering whether this will be merged into mppenc itself at some point?
A riddle is a short sword attached to the next 2000 years.

psymodel 1 and 2 of toolame (.mp2)

Reply #7
Psymodel 1 was developed by MUSICAM group, for Layers 1 and 2 (MUSICAM - Philips & Co)

Psymodel 2 was developed by ASPEC group, for Layer 3 (MP3) (ASPEC - AT&T, FhG & Co)

Psymodel 2 is >better< than Psymodel 1


Like Frank said - Psymodel 2 also has problems, like bad tonality detection - but it is better than psymodel 1

psymodel 1 and 2 of toolame (.mp2)

Reply #8
So is it sure, that model 2 is better than 1 at bitrates from 128 up to 224 (or higher) kbit/s in toolame ?

All at stereo, not joint stereo, because of keeping surround information ?

And for music and speech ?

psymodel 1 and 2 of toolame (.mp2)

Reply #9
If I am not mistaken, MPEG Layer II (MP2) doesn't have M/S stereo, just IS stereo - so Joint Stereo is basically Intensity Stereo - I wouldn't recommend this, except for MP2 @ 128 kbps stereo.

So, configuration for 192 Kbps and above:

Psymodel 2
Plain stereo


-- Ivan

psymodel 1 and 2 of toolame (.mp2)

Reply #10
Yes, that joint stereo vs. stereo issue in mp2 for keeping surround is clear.

So I tried and found out that even 112 - 128 kbit/s in plain stereo in mp2 is a acceptable compromise for keeping surround.
With 96  kbit/s and lower  in plain stereo the sound will get very worse.

The reason why 112 or 128 kbit/s in plain stereo in mp2 is better compromise than 112/128 in joint stereo is:
this format is used for movies, so mostly no music, more speech, sounds and surround effects.
So stereo mode for keeping surround is more important than HighFrequ., exact tonality.


So do you recommend psy model 2 over 1  for low bitrates, 96, 112 up to 160 kbit/s, all plain stereo, too ?
Or should model 1 preferred for bitrates 160 and lower in plain stereo ?

psymodel 1 and 2 of toolame (.mp2)

Reply #11
I wouldn't recommend Model 1 anywhere

psymodel 1 and 2 of toolame (.mp2)

Reply #12
Can anyone direct me to some documentation (code also would be great) about Mpeg Layer II?

PM me if so. Thanks in advance.

psymodel 1 and 2 of toolame (.mp2)

Reply #13
Hi,

after it is confirmed that model 2 is better than 1, I wanted to try  it.

But:

toolame stopped before song was finished !

Win XP, GUI Besweet with batch encoding, using: ac3 to wav:azid,
wav48 to wav44,1: ssrc 1.29_hp, 2 passes, wav to mp2: toolame...

Using latest versions of all programs, available at www.doom9.org eg.

This stopping I remember that I had it earlier, too.
But since some time I had no problems at all, everytime using model 1.

Have just tried with same conditions model 1, and:
songs are encoded till the end, not with model 2, everytime same switches, programs.

STRANGE ?

The first song I encoded today with model 2, that one was encoded until end.
But all later ones not...


psymodel 1 and 2 of toolame (.mp2)

Reply #15
hm, do you think it is the front end ?

I don't, as Besweet is used by a lot people and it just passes parametres to exe-files.
Has somebody errors with toolame model 2, too ?

The advantage of Besweet is, that in batch encoding the best programs azid1.8, ssrc1.29_hp and toolame12i are used, one step ac3 to mp2.
There are intermediate files, after ssrc processing there is one "big" wave.  toolame is started properly, it shows correct parameters in DOS-box.
Eg. the first 4 minutes of a 20 minute wave get encoded, no more.
Several trials on the same ac3/wav result to the same mp2, which is cut exactly at the same time.

Just changing in Besweet psymodel from 2 to 1, then the whole song becomes encoded.

psymodel 1 and 2 of toolame (.mp2)

Reply #16
User, I've found that this one has been the most stable using psymodel 2:

http://www.macdvd.org/trnsz/Compiles/TooLAME-0.2i-Win32.zip

hope it helps!

OnkelT
"Those are my principles. If you don't like them I have others".
                      Groucho Marx (1890-1977)

psymodel 1 and 2 of toolame (.mp2)

Reply #17
My first trials are successful with that  .exe !!!

many thanks. I would suggest, that at doom9 the download section should get updated with this version.

psymodel 1 and 2 of toolame (.mp2)

Reply #18
Glad that I could help. I've had exactly your problem user, the DOS version just stopped encoding large wav's...

About psymodel 2 - why does the VBR mode produce significantly smaller files when using psy 2 than psy 1? Is this due to a more effective encoding or is it because psy 2 isn't tuned for VBR?

Thanks Speek for your TooLame frontend 

OnkelT
"Those are my principles. If you don't like them I have others".
                      Groucho Marx (1890-1977)

psymodel 1 and 2 of toolame (.mp2)

Reply #19
Quote
Originally posted by Ivan Dimkovic
If I am not mistaken, MPEG Layer II (MP2) doesn't have M/S stereo, just IS stereo - so Joint Stereo is basically Intensity Stereo - I wouldn't recommend this, except for MP2 @ 128 kbps stereo.

So, configuration for 192 Kbps and above:

Psymodel 2
Plain stereo


-- Ivan


MPEG Layer 2 has Mid/Side Stereo. Can be switched frame based.
Intensity Stereo can be switched frame based from a selectable
frequency. Frequency can be selected from fs/8, fs/4, 3fs/8, fs/2.
--  Frank Klemm

psymodel 1 and 2 of toolame (.mp2)

Reply #20
Frank,
if I understand the ISO/IEC specification correctly, MP1 and MP2 really only knows IS.
Only MP3 allows MS coding.

psymodel 1 and 2 of toolame (.mp2)

Reply #21
Yes, Frank - me, too - I think that Layer I and II have IS only

And, that Andree introduced M/S in MP+ by inserting M/S similar to AAC into MP2 coding kernel?

 

psymodel 1 and 2 of toolame (.mp2)

Reply #22
On my copy of the standard, mp2 doesn't have M/S neither, only IS