If you need a stop-gap solution while you're waiting for audacity, you shou=
try out ecasound (http://www.eca.cx/). It's a sweet little program. =20
Actually, I prefer to do my recording with it, and mix with Audacity. Not=
that I've had much time for fun stuff lately... :(
Besides that, I recall others having trouble with full-duplex recording und=
certain Mandrake versions with certain sound cards, but I don't recall=20
exactly where it went. It was right before Josh rewrote AudioIO a few mont=
On Thursday 01 May 2003 08:09 pm, Hisham H.Muhammad wrote:
> I'm aware that full duplex on Linux is still problematic, and I'd like
> to know if someone is actively working on this. If not, I'd like to try
> to tackle this problem. I spent some hours today reading the sources and
> verifying the values of variables during program execution. I'm focusing
> on AudioIO.cpp and the files at portaudio. I'm working on the sources of
> Audacity 1.1.3 (couldn't compile today's CVS, but apparently the sources
> of AudioIO.cpp and the portaudio directory are still identical).
> At this stage, I'm still getting acquainted with the code. I can mostly
> understand what the code is supposed to do; the next step is to find out
> what it is actually doing. :) Still, there are some things I couldn't
> figure out by myself yet:
> * there are some #defines throughout AudioIO.cpp, for PORTMIXER and
> USE_PORTAUDIO_V19. In 1.1.3, none of these are defined. Should I replace
> PortAudio with version 19 (which I suppose is the CVS version, since
> latest in portaudio.com is still 18) and base my work on it? Anyway, I'm
> pretty certain that the problem is not with PortAudio. The example
> patest_wire in pa_tests does full-duplex and works fine.
> Actually I only started to believe my card can do full-duplex (and hence
> pursuing getting Audacity to work) after I found out GNUitar works well
> with full-duplex. The only ways I got full-duplex to work on my system
> until now were (1) with GNUitar, (2) with the patest_wire example, and
> (3) playing with 'play' (the sox wrapper) and recording with Audacity.
> Interestingly, playing with 'play' and recording with 'rec' causes a
> very similar problem from that caused by Audacity: the recording sounds
> one octave lower, albeit not sliced in 1.5sec (128KB?) periods as it
> happens in Audacity.
> This leads to the possibility that it's all just a matter of initializing
> reading and writing modes correctly. Both GNUitar and patest_wave
> perform open() in RDWR mode (but, based on (3) we know there's a
> combination with two open's that also works). I noticed that GNUitar
> calls a SNDDSP_DSP_SETDUPLEX ioctl, and PortAudio doesn't. I tried
> adding that but no luck (this was before finding out about patest_wire:
> now I know PortAudio can work well without this ioctl).
> * Another issue I couldn't understand clearly yet about the functioning
> of audio capture in PortAudio+Audacity is how sample formats are
> managed. Apparently PortAudio (at least the OSS version, haven't checked
> the others) internally uses 16-bit capture only, right? It calls
> Pa_CallConvertInt16 to convert the data to the format requested by the
> application in order to feed the callback function. I must be missing
> something trivial here (since this same code works in half-duplex mode),
> but what confuses me is that, tracing the execution, the switch() on
> sample format selects paInt16, which does not perform any conversion,
> sending the data directly to audacityAudioCallback. Inside
> audacityAudioCallback, however, the data (inputBuffer) is casted
> straight into a float pointer. What's happening there?
> Well, this is what I could gather up to this point. I'm still kinda lost
> but I _really_ want to get this working. My only other option, as it
> seems, would be installing an additional operating system in my computer
> in order to use a single application, something I do not want to do.
> Ah, for reference, the soundcard I'm using is my notebook's onboard
> Yamaha OPL3SA2, using ALSA 0.9.2 drivers with OSS emulation. If I
> fail getting OSS to work, I might end up attempting a port of PortAudio
> to ALSA/JACK.
> Hisham Muhammad
> S=E3o Leopoldo, RS, Brazil
> This sf.net email is sponsored by:ThinkGeek
> Welcome to geek heaven.
> Audacity-devel mailing list
"I slipped inside the oval office,
I slipped in oh so fast,
Grabbed the president by the necktie
And wiped my funky ass, hey"