Help - Search - Members - Calendar
Full Version: What is the best MP3 setting for encoding speech?
Hydrogenaudio Forums > Lossy Audio Compression > MP3 > MP3 - Tech
Rizban76
I have some audiobooks on CD that I'd like to encode so I can play them in my portable player. I could of course use -aps but I thought that there might be some ideal settings for encoding speech.
Does anyone have any tips for encoding speech with LAME?
neoufo51
Use the presets included with RazorLame.

But if its just speech, 20-32 kbps VBR at 22050khz will be enough.
floyd
I have a bunch of audio tapes at mono 22khz 32kpbs cbr, and while acceptable, the mp3 sound grates on you after awhile. I'd go mono 22khz 48kpbs or drop the bitrate down lower and use 11khz.
Old Nick
What would be the best setting for 32 khz, Mono? I'd like to encode some audio books, but my mp3-player only accepts MPEG1-Layer3, so 22 khz is out of the question.
/\/ephaestous
if you must use Mp3 AFAIK the Fraunhofer codec is better for lower bitrates. If you don't have to use mp3 then use speex.
Old Nick
But isn't the Fraunhofer codec better only for low bitrate STERO-files?
floyd
QUOTE (Old Nick @ Mar 8 2003 - 07:50 AM)
What would be the best setting for 32 khz, Mono? I'd like to encode some audio books, but my mp3-player only accepts MPEG1-Layer3, so 22 khz is out of the question.

what? are you suggesting mp3 doesn't support 22khz? works fine on my creative nomad II.

edit: it appears you are right. I did some searching and indeed mpeg1-layer3 doesn't support lower than 32khz. In that case I'd probably use lame or fhg at around 48kbps for 32khz mono audio. Also, setting the lowpass to 10khz or something close could help alot.
M
QUOTE (floyd @ Mar 8 2003 - 04:51 PM)
I did some searching and indeed mpeg1-layer3 doesn't support lower than 32khz.    In that case I'd probably use lame or fhg at around 48kbps for 32khz mono audio.  Also, setting the lowpass to 10khz or something close could help alot.

Probably not necessary to cut things off quite that low, as most audiobook programs shouldn't suffer from a slightly extended spectrum. But here's a fairly reasonable line, if you would like to try it:
CODE
--alt-preset cbr 48 -a --lowpass 15 -t

In this example, L.A.M.E. matches the default cutoff frequency for --alt-preset cbr 96 (stereo), downmixes and automatically resamples to 32kHz. That's nice, since a few audiobooks include written songs, and if the reader chooses to sing them (Robert Inglis' dramatic readings of Tolkien's The Hobbit and Lord of the Rings trilogy spring to mind) the more generous cutoff will allow a little musical "breathing room."

For audiobooks I prefer CBR, but this is an arbitrary decision... leave it in or out, as you please. (I'll only note that in some cases, despite the fact that all MP3 players should support VBR, CBR is a more reliable option.) Also, since I object to the inclusion of extraneous tags (unrelated to the program material, that is) I use the -t switch to turn off the L.A.M.E./Xing tag. This is generally frowned upon if you are trading MP3s; I am not, so I don't particulary care that it's frowned upon. Again, do as you please.

- M.
Dibrom
http://www.hydrogenaudio.org/forums/index....e,and,speech&s=

Please use the search function...
Old Nick
Thanks M, I'll try that commandline. smile.gif
This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please click here.
Invision Power Board © 2001-2009 Invision Power Services, Inc.