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: warm/attack loss due to lowpass filter? (Read 3714 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

warm/attack loss due to lowpass filter?

i experienced some warm and attacks losses when playing on my personal hi-fi a encoded/decoded/burned (lame aps) comparing to the original disc...

is this due to lowpass filtering?

when analysing frequencies of an 'aps encoded' file, i can see the behaviour of the athcurve which prefer cutting or smoothing high frequencies over 16khz to give more bit the to middle ones... (am i wrong?).
this makes (to me) guitar harmonics and alto sax sound a bit different (like smoothed...).

can switching from transform coder (mp3) to subband coder (mpc) help this?

what about the -k switch with lame aps? (:eek: ?)
does mpc cuts high frequencies off? with similar quality/size ratio than lame aps - using mpc --extreme?

please give me light on this (i'd like to have dibrom's or mithrandir's opinion on this...).
and scuze but i'm french and i have difficulties to express this in english...

cu,
pipo.

warm/attack loss due to lowpass filter?

Reply #1
No, it's not lowpass filtering, but what we call pre-echos, something which plagues transform coders, even at very high bitrates. Switching to a subband coder like MP2 or MPC will help. Changing frequency cutoffs will not help.

warm/attack loss due to lowpass filter?

Reply #2
Generally speaking mpc --xtreme "cures" anything... so I guess switching to mpc would be a good idea...

Frequency analyzing isn't a good method to say if a encoder is good... mpc has (as far as I know) an adaptive lowpass, so you can end with HF cut at 17khz or at 21khz if the HF energy is high enough (the lowpass is up to the psychoacoustic implementation)

Anyway, mpc is very well tuned so with mpc --xtreme you can't go wrong at bitrates ~200kbps.

warm/attack loss due to lowpass filter?

Reply #3
As others have already pointed out, this "loss of attack", is probably pre-echo.  In that case, yes, switching to MPC would effectively eliminate this problem.

As for --aps cutting out frequencies over 16khz, it doesn't do this so much by design.. instead it is due to a limitation in the MP3 format which makes encoding these frequencies very expensive.. and to keep bitrates reasonable, often times the encoder must refrain from scaling encoding accuracy too high (which it would attempt to do to encode these).

Using the -k switch with the --alt-presets isn't recommended.  Not only will it increase the bitrate significantly, it could also degrade the quality.

MPC doesn't cut out high frequencies to the extent LAME does.. and in fact, doesn't actually use an adaptive lowpass filter, though the behavior it exhibits is similar.  Instead, it cuts out frequencies according to the ath and other calculated masking information, and then just decides whether or not to encode that subband (which seems similar to lowpassing, though it's not quite the same).  At any rate, the ath used in MPC --xtreme is very low, and usually will encode content up to around 20-21khz or so on a fairly regular basis (very rarely dipping below 18khz I believe).  In short, you should have no problems with this.

warm/attack loss due to lowpass filter?

Reply #4
yes, i know frequency analysis don't say about quality, i just wanted to see what frequencies were kept...

as i was browsing the board, i saw different opinion about mpc 'transparency':

some use: --xtreme -nmt 16 -tmn 32 (very high bitrate, damned...)
dibrom you said you were adjusting the noise masking to 9 or 10 to get more transparency: --xtreme -nmt 10
others use: --xtreme or --standard (just alone)
what about (i conclued to this  may be just like extreme with lowered tmn): --standard -ltq fil -nmt 10?

question would be:
as i may burn albums to cdr a day or another, must i use 'transparency' or 'archival' presets, keeping an average bitrate below 250kbs? what are the advised switches to?

dibrom, are you expecting to provide --alt-presets or a list of recommended settings for mpc (but i guess mpc doesn't need special presets... :rofl: )?

sanx again...

pipo.

warm/attack loss due to lowpass filter?

Reply #5
Quote
Originally posted by don_pipo_corleone
yes, i know frequency analysis don't say about quality, i just wanted to see what frequencies were kept...

as i was browsing the board, i saw different opinion about mpc 'transparency':


Keep in mind what differentiates "opinion and preference" from "proven necessity".

Quote
some use: --xtreme -nmt 16 -tmn 32 (very high bitrate, damned...)


This has already been discussed, but so far, nobody has proven that they need this high of a setting to reach transparency on a regular basis.  It's way overkill.

Quote
dibrom you said you were adjusting the noise masking to 9 or 10 to get more transparency: --xtreme -nmt 10


Yes, but this was only on a very limited number of cases, and only on music which has proven to be extremely difficult to encode for almost all other encoders.  This type of music would be considered as a "codec killer".

Quote
others use: --xtreme or --standard (just alone)
what about (i conclued to this  may be just like extreme with lowered tmn): --standard -ltq fil -nmt 10?


I'd just use --xtreme since it would be tuned better than that.

Quote
question would be:
as i may burn albums to cdr a day or another, must i use 'transparency' or 'archival' presets, keeping an average bitrate below 250kbs? what are the advised switches to?


Unless you can actually differentiate (and prove it to yourself) that you can reliably hear a difference between --xtreme and the original (or even --standard for that matter), I probably wouldn't use anything higher.

Quote
dibrom, are you expecting to provide --alt-presets or a list of recommended settings for mpc (but i guess mpc doesn't need special presets... :rofl: )?


I doubt it... because as you said, MPC doesn't need something like that.  The default presets already work perfectly.

warm/attack loss due to lowpass filter?

Reply #6
sanx!