Category Archives: Languages

Approaching understanding

Leave a reply

Today was a good day of slow progress! Here’s the play by play.

  • Building a copy of XAudio2BasicSound, cleverly calling it ConsoleSound2.
  • Since it depends on SDKwavefile, I had to pull SAFE_DELETE, SAFE_DELETE_ARRAY, and SAFE_RELEASE from dxut.h. It also requires dxerr.h. but that’s in the regular directx include directory.
  • Extremely empty code right now, but it compiles and links. On a side note, the compiler should accept the environment variable $(DXSDK_DIR), but it doesn’t. And adding macros is an enormous pain, as near as I can tell.
  • The code is ported over and works. It’s all stuck in a main() right now. I need to clean up and build an Audio class. Actually, looking at the way the code is set up, a WavSampleSound class might be better. It probably could extend from a SampleSound class, but I’ll break that out later…
  • Class is mostly done and running!
  •  ConsoleSound2
  • The only bug that I have is that I can run the sound only once. I still need to figure out how to reload the buffer.

3D sound compiled and running, if not understood all that well…

Leave a reply

Today’s twists and turns:

  • More Audio3d.
  • Starting to add in the 3dAudio based on this and this and this.
  • Got an audio link error and chased that down, discovering a few things along the way.
  • There is directX help for C++ in (“C:\Program Files (x86)\Microsoft DirectX SDK (June 2010)\Documentation\DirectX9\directx_sdk.chm”). I don’t know it it’s any good, but I’m going to look though it next. Note that this is DX9, not 11. I’m *hoping* that this is all up to date…
  • Pointed Configuration Properties->Linker->General->Additional Library Directories at (C:\Program Files (x86)\Microsoft DirectX SDK (June 2010)\Lib\x86), and added the following files to Configuration Properties->Linker->Input->additional dependencies”
    • d3d11.lib
      d3dx11.lib
      winmm.lib
      xinput.lib
      X3DAudio.lib
    • This is a change from the libraries that the demo book points at, which were dxd9 libs. Recompiled and ran the code with no problems. I did attempt to clean up the xact3dInstance pointer in ~Audio3d(), but delete[] threw an exception
  • Reading through the DX documentation, which, though it is in the DX9 folder, contains information about DX11. Whatever.
  • The documentation is kind of what you’d expect, but it let me find the tutorials and demos, which are quite nice. On my machine, the demos that I’m interested in are located here: (C:\Program Files (x86)\Microsoft DirectX SDK (June 2010)\Samples\C++\XAudio2). It contains 4 projects:
    • XAudio2BasicSound
    • XAudioBasicStream
    • XAudio2CustomAPO
    • XAudio2Sound3D
  • Each project has source and MSVC2008 and MSVC2010 project files. I’ve just opened up the XAudio2Sound3D MSVC2010 project, and it compiled and ran just fine: XAudio2demo
  • This is almost exactly what I intend to build for the test environment, so that’s pretty cool. And using the buttons or the keyboard, I can drive the sound emitter around my head. Woohoo!
  • Next goal is to build up a copy of XAudio2BasicSound and then add the 3D components (without the graphics) from XAudio2Sound3D.

3D sound and the search for a header file.

Leave a reply

Today, I was going to extend the audio app to do some simple 3D sound. Nope. Aside from a lot of distractions that really only allowed me to get a couple of hours work done, the C++ documentation for this capability is getting harder and harder to find. And for some reason, Microsoft does not seem to believe in producing samples of code that actually do something. Instead, there are lots of nested pages, each pointing to a tidbit of information. Sigh. Anyway, here’s today’s struggles.

  • I thought this tutorial would be the best way to get positional information, but it appears to be stale. The New Way seems to be this.
  • Close, but it’s actually part of the Windows API now?
  • Nope that’s C#. Still looking. Since I have xact3.h working, trying to stay in this space…
  • Ok, at least for now (and yet again!) I think I know what’s going on and what I should do next. For tomorrow, I start writing a 3D version of Audio.h/Audio.cpp and then try calling that from the ConsoleSoundApp.

A much better day…

Leave a reply

Today’s stuff.

  • Imported the sound example code from P2DG chapter 7 into MSVC and successfully compiled.
  • And runs! With sound!
  •  withSound
  • Next we try to put the audio class into a console app (main as opposed to winmain) and see if that can be made to work.
  • That went pretty smoothly. I had to add the DirectX “include” directories, add stdafx.h to the audio.cpp file and cast some char* to LPWSTR. The next step is to fill out the main() so that the audio is initialized and then run a loop to send different audio commands. A good task for Monday.

Beating on OpenAL, then giving up. Microsoft wins.

Leave a reply

Progress for today

  • Copied the OpenAL SDK out of the ProgramFiles(x86) directory where the installer put it and put it in my menagerie of API test code
  • Converted the MSVC 2005 project to 2010
  • Tried to compile, but the target names didn’t match. Adjusted the output name in the Properties->Configuration Properties->Linker->General->Output File field from PlayMultiChannelWin32.exe to PlayMultiChannel.exe. 
  • Compiled and ran in debug mode, but got the same problem where the audio is going to the 3 front speakers.
  • Since all else has failed, I’m going to RTFM. In OpenAL 1.1 SDK/docs is a collection of documents. Going to start with the OpenAL Programmer’s Guide.
  • So this is really weird. I’m looking through the code to see how the channels are activated, and there doesn’t seem to be anything explicit. But I do find where the wav files are stored. Listening to these samples in windows media player, the correct Dolby channels are selected for 5.1 and 7.1. It seems as though OpenAl is flattening the information to the front three channels. Wile poking around on the OpenAL website, I came across the following regarding Vista:
    • With Microsoft’s decision to remove the audio hardware layer in Windows Vista, legacy DirectSound 3D games will no longer use hardware 3D algorithms for audio spatialization. Instead they will have to rely upon the new Microsoft software mixer that is built into Windows Vista. This new software mixer will give the users basic audio support for their old Direct Sound games but since it has no hardware layer, all EAX® effects will be lost, and no individual per-voice processing can be performed using dedicated hardware processing.
    • I think I’m getting “basic” audio support. This explains why the codebase has not been touched since 2008 or so. Looks like a dead end.
  • Moving along, it’s time to try DirectX. The samples I can find online don’t really thrill me, but looking at books on Amazon, I found Programming 2D Games (P2DG), which has a nice section on sound. It’s now sitting in my Kindle Library. Let’s see how that goes.
  • Downloading the DirectX SDK. All 572 MB of it. Last updated in 2010. Why is this making me nervous?
  • Ran the DirectX installer, which blew up, but seems to have installed all the components, and the demos run. The demo games do generate Dolby 7.1 as well, so that’s progress…
  • Following the steps in Chapter 7 (Sound) in P2DG. XACT opend up just fine. Importing wav files from the OpenAL media directory. Interestingly, some .wav files choked as “unrecognizable” Don’t know why…
    xactError
  • Got the sounds imported and playing in XACT and the Audio Console. And considering how much stuff didn’t work today, That’s enough progress to leave on a high note.

A bit of a new direction

Leave a reply

Somewhere in the middle of finals, I realized I had this nice FLTK/OpenGL application I’d written in C++:

It currently calculates and outputs audio commands to an Arduino that in turn talks to MIDI boards for each channel. But if I can use the main CPU’s internal audio capability, then I can have a deeper level of capability, and it works for enough of my current projects that it’s probably worth building some generic capability.

So that means I need a C++ library and API that can be built using the MSVC 2010 IDE. Possible options are:

Standard programming mysteries.

Leave a reply

I hate it when I’m starting off with an API and the tutorials don’t work.

I was working with the LWJGL OpenAl examples described in earlier posts, and could not get the audio files in the examples to open. The file was there, it could be converted to an InputStream, but when I tried creating a LWJGL WaveData object, the underlying java audio code would throw an error and I’d get a null object.

So since I’m now trying to figure out what gives with the javax.sound.sampled.* system, I might as well see if I can do what I want to do here. So off we go to the Internet once more, looking for examples that work.

The first thing I found was the Oracle Java Sound API documents, which are huge, loaded with theory and very little code. I’m not complaining really, because they look like they will be really useful when I have a little bit of code working. But there needs to be a “Hello, World”.

Which showed up after a bit of searching. There is a website JSResources, put together by two German (?) developers; Florian Bomers and Matthias Pfisterer. It is loaded with examples that seem to work. I’ve just gotten one of their simple examples running with a minimum of fuss. (Here’s the (slightly) modified code).  Since this is straight sound library, we know that we can run all the channels. Now we just need to get from here to there.

Multichannel sound

Leave a reply

It was looking like Java3D only handled Mono and stereo, but I just found com.sun.j3d.audioengines.javasound .JavaSoundMixer, It also uses javax.media.j3d.PhysicalEnvironment, which looks like it has built-in head tracking.

Darn it – got my hopes up too soon: “While the sound image created for final output to the playback system is either only mono or stereo, most device driver implementations will mix the left and right image signals generated for each rendered sound source before outputting the final playback image. Each sound source will use N input channels of this internal mixer. Each implemented Java3D audio device driver will have its own limitations and driver-specific characteristics. These include channel availability and usage (during rendering). Methods for querying these device-driver specific characteristics are provided.” (from javax.media.j3d.AudioDevice)

OK, I’m not ready to give up on Java yet. According to this bug report, multichannel support has been working for a while. My two options at this point are OpenAl through the LWJGL folks, and the javax.sound.sampled library.

Because I’d like to show graphically where the sounds are coming from, I’m going to try the LWJGL approach first.

Looks like multichannel support is in OpenAl, but will require care: “The multi-channel extension provides a mechanism to play multi-channel data via OpenAL. A variety of formats are supported. Multi-channel buffers can be attached or queued on a source. Note that when using the “Generic Software” device, the multi-channel buffers are mixed down to a stereo output. On a hardware device (such as the “Generic Hardware” device or a native device), each channel of a buffer requires a hardware voice. So, for example playing a buffer using the AL_FORMAT_51CHN16 format will require 6 free hardware voices. If the hardware resources are unavailable, the call to alSourceQueueBuffers or alSourcei will fail.”

And then there’s Javascript….

It turns out that there is a good deal of work going on in bringing multichannel audio to your browser. Since this *is* the future, it’s not a thing that I want to overlook, but I don’t want to put all that much time into it right now either. So here are some links:

This looks interesting

Leave a reply

First, I got the headset. Very nice.

Second, there is a company, sixense, that looks to be making some very good immersive hardware and a free(?) api. This might be very good for the pointing test. And they have some nice sound (midi) code too. It’s somehow tied up with Intel’s perceptual computing effort. I learned about this from a slashdotted article about the Holodeck Project.