Re: Enhanced PLAY.CMD with play list support

From: Michael Baldwin Bruce (mbbruce_at_mighty.co.za)
Date: 11/05/04 

Date: 4 Nov 2004 16:53:30 -0800

bruce aka tholen@antispam.ham wrote in message news:<Epfid.61086$Kl3.53120@twister.socal.rr.com>...
> Marty writes:
>
> >>>>> Peter Weilbacher wrote:
>
> >>>>>> Michael DeBusk wrote:
>
> >>>>>>> Marty wrote:
>
> >>>>> The most important aspect of the script for me was the resource leak
> >>>>> fix. Something inside of Timidity_MCD does not get cleaned up correctly
> >>>>> without the call to mciRxExit(). Strangely, missing this call doesn't
> >>>>> seem to cause a problem with any other codecs that I've found, just
> >>>>> Timidity. I guess the other codecs are smarter about cleaning up after
> >>>>> sloppy processes.
>
> >>>> I've probably encountered that problem. There have been times when
> >>>> PLAY.CMD would fail, while other audio file playing utilities would
> >>>> continue to work, including some Windows-based ones. But I have also
> >>>> encountered one instance in which PLAY.CMD worked from inside my editing
> >>>> program but failed as a standalone utility in another windowed command
> >>>> prompt session. Haven't been able to figure that one out.
>
> >>> Did you get any messages from PLAY about a "REXX Alias" or something to
> >>> that effect?
>
> >> Unfortunately, I cannot reproduce the problem on demand, so I have to
> >> rely on memory for the error message, but it was something like
> >> "mciRxInit routine not found".
>
> > I've seen this type of thing before with other DLLs (most prominently
> > REXXUTIL). The call to SysLoadFuncs returns a success, but REXX later
> > claims that the functions which were allegedly just loaded are not
> > available. I think this is a separate issue from the ones that I
> > addressed with my changes.
>
> A separate issue involving a bug in REXX?

Who's Rexx, bruce?

>
> >>> Instead of maintaining handles with binary data and the
> >>> like, in REXX you create a name or "alias" to associate with your MCI
> >>> device. Every multimedia operation you do in REXX is done against this
> >>> alias so that the state is maintained by the MMPM libraries, but your
> >>> context is maintained.
> >>>
> >>> Unfortunately, PLAY.CMD hard-codes a REXX alias name, so that if you
> >>> attempt to run two of them at the same time, the same alias is used and
> >>> the context is applied from your original running PLAY.CMD. Likewise,
> >>> if you fail to close out and free the REXX alias (due to killing the CMD
> >>> process, for example), its state continues to be maintained by the
> >>> system. The alias is global. You have to free up the REXX alias before
> >>> it can be used again.
>
> >> Well, that seems consistent with one occasion in which I was repeatedly
> >> playing a very short audio clip, pressing the key to repeat the play
> >> very rapidly, such that it seems possible the system started up the next
> >> PLAY before completely cleaning up after the previous PLAY.
>
> >>> My version of PLAY checks for this condition and performs the free if
> >>> needed.
>
> >> I'm not sure under what conditions my audio subsystem supports the
> >> simultaneous playing of two audio streams. I'm using the SoundMax
> >> audio built in to the Intel 850 motherboard (ADI 1885). I know that
> >> if I'm playing a .wav file, then fire up Mahjongg, it complains about
> >> the device being busy, even though I've disabled sounds in Mahjongg.
>
> > Even if your audio drivers are capable of audio stream mixing, an
> > application can still open the device in exclusive mode. What are you
> > using to play the wave file?
>
> I've tried a variety of different utilities over the years. I probably
> use PLAY.CMD more often than anything else. I like the simplicity of
> the command line interface.
>
> > I don't think the system will default to
> > opening it exclusively if you double-click a wave file or use PLAY.
>
> Well, Mahjongg complains that the device is locked when PLAY is
> being used to play a .wav file. When it's finished playing, no
> such message occurs. Maybe Mahjongg wants exclusive access and
> refuses to share, but then it shouldn't say that the device is
> locked when it really isn't (if that is indeed the case).

lol! Silly! Mahjong is a Chinese game of tiles. It can't complain
about anything.

>
> >> On the other hand, the driver came with WarpMix, so it looks like it
> >> should be able to mix at least a few audio sources.
>
> > I think WarpMix is tool used to set volume levels of various audio
> > channels (CD, line in, etc.), correct?
>
> Yes.
>
> > That's different from stream
> > "mixing" where different audio streams are combined, in software or
> > hardware in some cases, into one stream that is played. The former
> > process is an analog one and only happens among a fixed set of channels.
> > The latter is digital and can take place over a variable number of
> > audio streams.
>
> Odd to use "mix" in the name if it doesn't do any mixing.

Like Tom Mix, bruce?

>
> >>>> And I've
> >>>> also seen cases where the first invocation of PLAY.CMD would return an
> >>>> error message but a second invocation would cause the entire windowed
> >>>> command prompt session to die.
>
> >>> This one sounds like the MM libraries ran out of resources somewhere or
> >>> were left in an unstable state. This can be caused by not cleaning up
> >>> properly. I've seen this happen on my own system but haven't been able
> >>> to nail down the exact spark that triggered it. Since modifying
> >>> PLAY.CMD, I haven't seen this happen, but I'm not convinced that the
> >>> problem is gone yet.
>
> >>>> I'm curious as to how much digging others did to try and find the
> >>>> kinds of TIMEFMT supported by PLAY.CMD. I know it supports "samples"
> >>>> and "ms" (for milliseconds), but the default units are something like
> >>>> 15 samples (for 44.1 kHz audio). Haven't dug enough to understand
> >>>> the origin of that choice. Any other time formats supported?
>
> >>> 15 samples / 44100 samples/sec = .00034 secs of audio
> >>>
> >>> Seems like a strange choice indeed. It is approximately 1/100th of a
> >>> normal process time slice (32ms). That's a bit of a stretch, though.
>
> >> Sufficiently odd that I never use the default (other than to
> >> investigate it), which means I always have to use the TIMEFMT keyword,
> >> which is a bit annoying. On the other hand, my editing program works
> >> with the audio samples in memory, so to play an edited clip using
> >> PLAY.CMD, I need to write it to disk, so as long as I write just the
> >> clip I want to hear, I don't have to bother with the FROM, TO, and
> >> TIMEFMT keywords. Would be nicer to simply play a clip of audio
> >> samples directly from memory and thus avoid hitting the disk every
> >> time I want to repeat a short clip, but then I'd have to learn an
> >> entirely new part of the multimedia system. One of the editing
> >> tools plays a short clip whose end points I can move in 0.01 sec
> >> increments using the cursor keys; it has to write a new .wav file
> >> every time I press the key so as to utilize PLAY.CMD.
>
> > The learning curve for the multimedia subsystem proper can be quite
> > steep. It would allow you the ability to play from memory, volume
> > control, and data format independence (you could use codecs to read and
> > write your raw audio data), but getting there is tricky.
>
> Reading .wav files is sufficiently simple that I can avoid the overhead
> of a codec. On the other hand, handling every possible data type is a
> headache for the source code. There are digital audio units becoming
> available that can record 24-bit samples. Judging from the recording
> time available, they're not wasting space by padding the samples to
> 32-bit. If I wanted to handle files like that, I'd need to rewrite
> the code to read three 8-bit samples, because there is no native 24-bit
> data type.
>
> > In what language is your main project written?
>
> I rely heavily on a library of routines that I've written over the last
> two decades, largely in Fortran. The Fourier transform is an example.
> A lot of the "magic" my editor can perform (fixing wrong notes, for
> example) uses the Fourier transform.

Fortran, bruce? lol!!!

>
> > I ask because I'm
> > trudging my way through the same territory right now in one of my C
> > projects, and maybe I can provide you with a simplified code snippet to
> > get the system going for you. I have DART and raw MMPM code from past
> > projects, but they were muddled down in the mechanics of synchronization
> > with a video stream, so they don't serve as good examples.
>
> They might make good examples to the WarpVision people. The only time
> I tried WarpVision, I never could get decent synchronization with the
> video, even with the adjustment keys provided.
>
> > Also, they
> > only dealt in pumping raw data out to the sound card. My new code will
> > need to be a bit more robust (being format-independent, for example).
>
> But API calls are API calls, regardless of the language used to call
> them.

do you often call apis, bruce?

Relevant Pages

  • Re: Enhanced PLAY.CMD with play list support
    ... The call to SysLoadFuncs returns a success, but REXX later ... >> Even if your audio drivers are capable of audio stream mixing, ... >> That's different from stream ... handling every possible data type is a ...
    (sci.astro)
  • Re: Digital reception problems just on BBC in Brighton
    ... Found 1 video stream. ... Found 1 MPEG audio stream. ... Closing starting GOP. ... Video stream 1 now starts at 00:00:01.188 ...
    (uk.tech.digital-tv)
  • Re: Creating a speaking tapi application
    ... Now I want to play that audio with speaker instead of recording. ... It is called terminals because when we talk about video streams ... earlier date that perhaps I could monitor the incoming stream and attempt ... CComPtr pEnumAddress; ...
    (microsoft.public.win32.programmer.tapi)
  • Tips for recording audio streams
    ... recording streaming audio from the Internet on a Linux box. ... You can use it on the command line or from ... This records the audio stream linked to in RealAudio format ...
    (comp.os.linux.misc)
  • Re: VOIP Softphone
    ... The application receives a data stream off the network. ... spaced out every 50 milliseconds which equates to 800 bytes of audio data. ... in my application by either dropping packets if they are to late, ...
    (microsoft.public.win32.programmer.mmedia)
Loading