[Stk] Bug in RtAudio or base class?

John Heenan john@auscyber.com
Thu, 21 Nov 2002 06:29:28 +1100


Hi Gary

> >>Regrettably the effects tcl GUI demo is still not working. Yes it did
work
> >>before driver upgrade.
>
> That is very strange.  The "effects" project uses a soundcard in
> duplex mode, so it is using both the input and output routines in
> RtAudio simultaneously (as opposed to the "play" program which only
> uses output routines and the "record" program which only uses input
> routines).  This is typically the most problematic behavior to get
> working.  I'm sorry I did not understand before that duplex mode was
> working ... that would pretty much have ruled out any problems with
> your soundcard driver.  What sort of error message do you get when you
> attempt to run the "effects" program?
>

I go no error messages. There was no response to any 'button' pushing.
However I have gone back to the old drivers by restoring an image of my W2K
system partition. Effects is now working again. However it is working as
before with 'static' as if the recording level is too high. The static does
not occur with silence, only with sound.


> >>What can be set from the system are two audio driver parameters for both
> >>recording and playback.  These are 'sample rate conversion quality' and
> >>'acceleration'. For both recording and playback sample rate conversion
> >>quality was left at 'good' (the lowest of three settings'. There are
four
> >>acceleration settings. These are
> >>None - Emulation only
> >>Basic Acceleration
> >>Standard Acceleration: use this setting to enable only standard
DirectSound
> >>acceleration features. (Recommended)
> >>Full acceleration
> >>
> >>I got best results with the non recommended setting: Basic Acceleration.
> >>
> >>The best results do not include echoing or reverberation but do
> >>include what superficially sounds like as if the recording level had
> >>been left too high. I cannot eliminate this by reducing the
> >>recording level.
>
> I'll assume that playback is working correctly for you now (and that
> the WvOut fix I sent last week now allows you to correctly read and
> play 8-bit files) ... please let me know if that is not the case.
>

No. Playback was not working correctly with the updated drivers and is not
working now with previous drivers. The term 'best results' is relative to
what were some very obtursive additional echoing or reverberation effects.

> What happens when you run the "io" example program?  That tests duplex
> mode.

I get the same 'static' effect as if recording level is too high. However
adjusting the recording level does not change the quality. However the
results are 'perfect' with the io programs from the rtaudio test suite (see
below).

> From what I understand, you are still having problems with the
> "record" example?  On my Windows system, there is an "Advanced" button
> below the Mic volume control.  When I click on that, there is a 20dB
> setting which can be checked/unchecked.  If you have that, have to
> tried turning that off?

With the older drivers there is no such advanced option. With the updated
drivers there was an advanced option which included a 'boost'. Checking this
on or off had no effect on the quality.

>
> If there remain problems, my next suggested step would be to download
> the RtAudio distribution and run the tests there:
>
>   http://www-ccrma.stanford.edu/~gary/rtaudio/
>
> Each test program allows you to set the data format (#define
> statements at the top of each), so we could better determine whether
> the problems you are having are coming from RtAudio or other STK
> classes.  If you're into this, let me know and I'll provide more
> explicit recommended tests to run.
>
> --gary
>
I have downloaded the test programs with the following results leaving the
data formats defines as is.

Unlike io.exe both call_inout.exe and inout.exe gave 'perfect' reproduction
although there was noticeable slight latency.

The effects of using record_raw.exe followed by play_raw.exe was not as
good. The 'static', as described above, was present. Adjusting recording
level had no effect.

John