MP3 Presets that Equalled FLAC
Dec 26, 2006 at 3:56 AM Thread Starter Post #1 of 9

EFN

Headphoneus Supremus
Joined
Aug 14, 2006
Posts
3,034
Likes
21
Yes, I have finally found the perfect LAME 3.97 presets that's performing equally to my FLAC VBR8 library.

And the presets are:
--clipdetect -V0 --vbr-new -b128 --lowpass 21 -q0

It's a VBR New with Joint Stereo and very accurate Replaygain Clipdetect on the fly.

I know I'll get flamed for this. I have been advocating FLAC all over the place and now I am saying that LAME VBR can come on par with FLAC. The fact remains that I am a hardcore FLAC fanboy, but recently my FLAC has been chomping on my DAP battery for breakfast and I am getting tired with frequent re-charging of my Rio Karma.

So I gave LAME 3.97 another shot. I have mentioned in a different thread before that basically a maximum quality LAME MP3 bear similar SQ to FLAC, my only gripe is that I can audibly hear very minor artifacts and clippings on peaky songs - so I abandoned my earlier VBRs in favor for a 100% FLAC setup. And please be reminded that I am using Etymotic ER-4S in all of my testings, no imperfections no matter how small can hide from this very revealing IEM.

2 days ago I played with LAME --clipdetect switch and I was pleasantly surprised by the results. I re-encoded around 430 of my favorite songs using the presets I mentioned above and sure enough I CAN'T detect not a single artifact/clipping on any of my re-encoded MP3s! This could only mean that LAME's --clipdetect is even superior to Sony Soundforge V8 Replaygain/Normalize which doesn't seem to be that efficient (and Soundforge is not free)

The only catch is that due to the permanent replaygain settings, my normal listening volume now has to be increased 2 notch up - I will not complain on this for as long as the MP3s remained clear and clean.

Cheers
etysmile.gif
 
Dec 26, 2006 at 6:13 AM Post #2 of 9
What program are you using to convert
Quote:

Originally Posted by EFN /img/forum/go_quote.gif
Yes, I have finally found the perfect LAME 3.97 presets that's performing equally to my FLAC VBR8 library.

And the presets are:
--clipdetect -V0 --vbr-new -b128 --lowpass 21 -q0

It's a VBR New with Joint Stereo and very accurate Replaygain Clipdetect on the fly.

I know I'll get flamed for this. I have been advocating FLAC all over the place and now I am saying that LAME VBR can come on par with FLAC. The fact remains that I am a hardcore FLAC fanboy, but recently my FLAC has been chomping on my DAP battery for breakfast and I am getting tired with frequent re-charging of my Rio Karma.

So I gave LAME 3.97 another shot. I have mentioned in a different thread before that basically a maximum quality LAME MP3 bear similar SQ to FLAC, my only gripe is that I can audibly hear very minor artifacts and clippings on peaky songs - so I abandoned my earlier VBRs in favor for a 100% FLAC setup. And please be reminded that I am using Etymotic ER-4S in all of my testings, no imperfections no matter how small can hide from this very revealing IEM.

2 days ago I played with LAME --clipdetect switch and I was pleasantly surprised by the results. I re-encoded around 430 of my favorite songs using the presets I mentioned above and sure enough I CAN'T detect not a single artifact/clipping on any of my re-encoded MP3s! This could only mean that LAME's --clipdetect is even superior to Sony Soundforge V8 Replaygain/Normalize which doesn't seem to be that efficient (and Soundforge is not free)

The only catch is that due to the permanent replaygain settings, my normal listening volume now has to be increased 2 notch up - I will not complain on this for as long as the MP3s remained clear and clean.

Cheers
etysmile.gif



 
Dec 26, 2006 at 7:42 PM Post #4 of 9
What's that "-b 128" doing in there?

BTW, the LAME docs say:
Quote:

--clipdetect clipping detection

Enable --replaygain-accurate and print a message whether clipping occurs and how far in dB the waveform is from full scale.


So --replaygain-accurate should have the same effect.

There should not be any improvement when using the higher LPF setting, not when using Etys at least (they reach like what, 16-17 kHz?).
 
Dec 26, 2006 at 10:00 PM Post #5 of 9
There's no reason to eschew lossy encoding for lossless in the first place. A 320 KB/s mp3 is identical to my ears next to the same song encoded with FLAC. The real point of lossless is archiving.
 
Dec 26, 2006 at 10:19 PM Post #6 of 9
Quote:

Originally Posted by sgrossklass /img/forum/go_quote.gif
There should not be any improvement when using the higher LPF setting, not when using Etys at least (they reach like what, 16-17 kHz?).


Exactly. Quality may be lower with a higher low pass if bits are wasted on the high frequencies. This is the reason why the low pass is set to what it is in the -V# presets.
 
Dec 27, 2006 at 2:16 AM Post #7 of 9
Quote:

Originally Posted by daveypadot /img/forum/go_quote.gif
What program are you using to convert


I use RazorLame Front End to encode (using custom presets only). Much easier. The actual Lame.exe still uses ver 3.97.




I have not been getting great success with 320kbps and other presets. The presets mentioned above seemed to worked best for me, others may differ so experimentation is necessary. But the key point is that LAME's "--clipdetect" does make huge differences in the output cleanliness.
 
Dec 27, 2006 at 2:31 AM Post #8 of 9
Quote:

Originally Posted by Altoids /img/forum/go_quote.gif
There's no reason to eschew lossy encoding for lossless in the first place. A 320 KB/s mp3 is identical to my ears next to the same song encoded with FLAC. The real point of lossless is archiving.


It's great for archiving, that's for sure, but with the advent of devices like the Squeezebox that's not all it is good for.

I use flac exclusively for listening at home.

But for portable use these new settings are very interesting.
 

Users who are viewing this thread

Back
Top