When time DOES matter

2003-03-05 16:32:51

For some time I started having a situation, when encoding speed is very important, but quality should not suffer as well.
I used to use --r3mix setting for quite a long time, as it provides near transparent quality at relatively low bitrates and encodes much faster, than APS. I don't have much complains on this mode, but in some cases (like well known applauds) it really sucked without any ABXing. IMHO even CBR192 kbps sounded better on this sample.
Another choice is using either 192CBR (very good quality, even for home stereo), but a couple of noticeable errors per album can drive me mad. Alt-preset insane is quite fast, but... it's really insane.
I haven't tested APS fast much yet. But is it worth using?
How do you think, what shall I use to get (near)transparent sound quality: --r3mix, APS fast or --insane? Encoding time really does matter.
P.S.: and, please, no answers like "try MPC". I'm using musepack for some months already and I'm completely happy with it. Just sometimes I need to make fast transcodes/encodes to play music on DVD.

When time DOES matter

Reply #1 – 2003-03-05 16:40:48

You'll find any of the --alt-presets, standard, extreme, or the fast versions, all sound better than --r3mix.

When time DOES matter

Reply #2 – 2003-03-12 02:20:47

also, i find lame 3.90.2 way faster than the 3.92 was ...

... and of course, API is very fast for encoding (way faster than aps/apx), but you might run into other issues there

alfred

When time DOES matter

Reply #3 – 2003-03-14 11:51:16

If you want faster encodes use --alt-preset fast standard.

P.S. Why -q 5 or -f have no effects? If you use them with alt preset it keeps using -q 2..

When time DOES matter

Reply #4 – 2003-03-14 16:21:03

You might also try Gogo encoder to see if it suits your needs....

I've played with it and it is very fast...Quality??  Well, it's based on an older version of Lame (3.88 I believe) so it's got good heritage anyway....It doesn't use the alt-presets.  If I remember correctly, it strictly uses the vbr settings (i.e. gogo -v X  where X is an integer from1 to 9)...

It might give you the solution you're looking for.  However, I'm not sure I can say it's consistently better than --r3mix at a given setting; I haven't really compared them....maybe others can comment on that.

Good luck....

When time DOES matter

Reply #5 – 2003-03-14 16:26:54

for transcoding i generally use the implementation of fastenc in musicmatch jukebox.  when running VBR it is fairly fast and seems to have good quality (expecially at lower bitrates... the kind you'd use on a portable)