My name's Bellegarde Laurent,
french teacher, using GNU/Linux only since year 2000, at this time under Ubuntu Feisty 7.04 or Ubuntu Studio 7.04.
I'm trying to create a documentation for making video tutorial for helping beginners using Free software as GNU/Linux and at the same time improving the use of multimedia tools under GNU/Linux with lprod.org free project, www.lprod.org.
my question :
is there a way to improve the audio bitrate of the ogg video file recorded by gtk-recordmydesktop ?
I'll get the audio track in 64 kb/s, which is a little bit bad for audio listening, 128 kb/s should be perfect.
which file should be edited to change this ?
Thank's for all, your software is great,
I'm not sure I understood exactly your problem, so let me know
if I'm offtopic.
Do you want to record audio with better quality or
increase the quality of an existing recording?
1) If you want better recordings:
The Vorbis codec uses variable bitrate to encode sound,
to maintain a costant quality. This setting is exposed in
recordMyDesktop as the -s_quality commandline parameter
and the sound scrollbar (scale widget) in the frontend.
Both have the maximum value by default.
If recording audio with the defaults doesn't give good
enough quality, you should change the samplerate of the
recording to 44100 or 48000 ( -freq 48000 in the commandline
and Advanced->Sound->Frequency on the interface).
The default is 22050, which might be low depending on the
quality you want to achieve. Of course having a good
microphone helps ;).
2) Upsampling an existing file:
This makes little sense as far as I now, especially if the
recording was made with the best quality settings offered by
the vorbis encoder.
thank's for your answer.
I'm using gtk interface, so i've put the sampling at the ax allowed 44100, 48000 is not allowed.
I'm using too 2 channels, the scroll bar is at 100%, so i think that's the best quality, and in theses conditions, the recorded video is 64kb/s audio bitrate, which is not good enough for listening as the humain hear feel the difference under 128 kb/s.
i want to improve the recording only.
after recording with recordmydesktop, i'm making a conversion into mpeg2 with ffmpeg, to edit it without downgrading the resolution with avidemux. The final video tutorial has 1280x1024 screen resolution, the images are excellent quality, but the sound is too low.
I'll hope i explain well that i want to improve.
>> I'm using gtk interface, so i've put the sampling
>> at the ax allowed 44100, 48000 is not allowed.
You found a bug there ;). It should allow higher values.
If you are using 0.3.5 or later you can override this setting
by entering -freq 48000 in the Advanced->Misc->Extra Options
But 44100 to 48000 shouldn't make much difference anyway.
>> the recorded video is 64kb/s audio bitrate, which is not good enough
>> for listening as the humain hear feel the difference under 128 kb/s.
The bitrate is a definite factor of quality, only when associated with
constant bitrate formats.
Like I said before, the Vorbis codec uses variable bitrate. Which means
that if your file manager or media player reports 64 kbs, it only
refers to the average bitrate.
When you are talking to the microphone, the bitrate is more
than 64 kbps, while whenever there's silence it can drop even more.
64 kbps is not a bad average for speech, as it will typically have
long pauses and it occupies a narrow frequency range (compared to
Now, beyond the theory, if the result isn't satisfactory to you
as a listener, I can think of two things:
1) The quality of the incoming sound isn't good enough.
Simple pc microphones are usually of a very low quality
and no matter how you compress the sound it will not sound
If the sound in the original ogg does not sound good, you should
look to improving the quality during the recording procedure
either by a) using better recording equipment or b) record sound
separately so you can edit it with something like audacity
and improve it's quality.
2) If you are transcoding to another codec, you might lose
a lot of quality. Vorbis is a lossy codec and transcoding to
another lossy one, like mp3, might render very poor results.
Especially when transcoding from a VBR (variable bitrate) codec
to a CBR (constant bitrate), you should make sure that the
new codec uses the maximum bitrate of the previous one and
not the average.
If you definitely have to transcode to another codec, then you
should use one that is also VBR (mp3 supports this).
My suggestion though is to use the original formats, as anything
else will result in at least some quality loss.
thank's for all the answer.
I think that 64 kb/s is the average, because i've compare the same recording done only with the same microphone, under audacity, and the end the sound seems to be the same.
After studing the audios controls, i've found a good recording.
I'm going to taste the encoding with a VBR codec.
At the end, the final result is not too bad in xvid test, you can have a look here :
I' think that's good now, thank's a lot.