I am a newbie to DirectX world
I am using latest DirectX SDK with Visual Studio 2005 and using C# as programming language I have checked the samples and they are working perfectly
I have an custom made external sound card (connected with USB) that has 2 inputs, 1 for left channel and 1 for right channel, DirectX SDK's Capture example is working perfectly with it capturing the sounds of both inputs and saving them in a wav file, now I want to save left and right channels' sounds in seperate files
I think it must be pretty simple but I have no clue about it
Thanks in advance
The data you got from directsound is interleaved. E.g. for each channel you get one sample, left channel first. All in one stream. That makes sense if you think about it: the two channels are sampled at the same time from the input, so they should arrive at the application at the same time as well.
In your case all you need to do is to split those two channels with a simple loop and write them to two mono wav-files instead of one stereo wav.
Thanks a lot your reply solved my problem
I too am trying to capture the stereo stream to two seperate wave files using the C# sample in the DirectX SDK. I am not quite sure what Nils Pipenbrinck means when he says "split those two channels with a simple loop".
Any assistance you can provide would be greatly appreciated.
The channels are interleaved. Like Nils says, there is a sample from the left channel, then a sample from the right channel, then another from the left, then another from the right, etc. You just write a loop that goes through pairs of samples and writes even-numbered ones to the left channel wave file and odd-numbered ones to the right channel wave file.
In other words. The left and right channels are multiplexed in the same sample stream. So you'll need to write a demultiplexer.
To visualize the multiplexed stream (each sample-set is wrapped in  and left and right channel are indicated by L and R respectively), so the incoming stream will probably look like this:
Each letter (L & R) here represents one sample. So all you need to do is loop over the entire input and split them out.
Keep in mind though that each sample is for example two bytes (assuming a system that has 8bit bytes here) long in case of 16bit samples.
my question is somewhat related to this one and goes like this:
I've written a 3D-Sound system that uses Direct3DSound using c++. I have EAX support and everything works fine.
BUT: This is a realtime app that allows users to move interactively but also save screenshots (even high-res ones by using many tiles that are combined later on) using predefined camera-paths and write out an avi file. The sound effects can be triggered by different actions like proximity, switches, ... . Since the high-res rendering sometimes cannot be performed in real-time (a lot of antialiased tiles have to be rendered offscreen and written to disk), I came up with a two-pass method for capturing sounds:
(1) only render the scene (be it tiled or not) to the disk and disable sound.
(2) turn on the sound effects, but don't render any visuals and follow the camera-path in real-time while capturing the sound-output using the "what u hear"-pin and write this to the avi-files sound track.
This works, but only with two-channels (left/right). I'd like to capture all channels (e.g. 5.1 sounds) and create avi's with that many channels using the WAVEFORMATEX format.
So the question is: How can I capture more than 2 channels? Is it somehow possible to capture each channel seperately?
Any help is appreciated, cheers,
Why do you need to capture the sound if your application is generating it? Can't you just compose the soundtrack by placing the sounds at the appropriate times and doing software mixing?
Thank you Reedbeta and MuggenHor. I now have my two seperate files, one left channel and one right channel by doing just as you explained.
probably I was not so clear when describing the problem:
The application (sort of a game engine, but different) is generating sounds/effects using DirectSound/Direct3DSound and EAX-effects for reverbation, echo, ... . The sound files are simple wav, ogg or mp3 files. These sound effects are placed in the 3d scene or linked to moving objects. An EAX environment is generated in a preprocessing step (analyzing the size of the different rooms, ...) and while the user navigates inside the 3d scene, the actual environment is determined to give the correct "ambient feedback".
The sounds themselves start/stop/loop according to the proximity of the virtual camera to these objects or by user-actions or by time, ... .
While this all works very good when achieving real-time frame-rates (like navigating in the scene) the problem is capturing the sounds (or soundtrack) to an avi file while rendering a lot of still frames. In this case the camera is moving along a predefined path and effects are triggered just like in the interactive mode. The camera movement is not real-time because of antialiasing techniques and the high resoulution. In my previous post I mentioned the workaround to this problem (sort of ugly but I can live with it), but I can only record stereo (2 channels).
So my questions are:
(1) Is there a way to capture the samples (like in the sampleGrabber-demo) in a one-shot mode (run sounds --> capture --> do_something_that_takes_a_long_time --> get next sample) while still getting EAX effects and more than two channels?
(2) If (1) is impossible, how can I capture the output of just one channel (say back-left). This way I could write each channel seperately to the combined avi-soundtrack.
(3) What exactly do you mean by software mixing? I know how to get hardware acceleration in DirectSound, but I thought switching to software mixing would decrease quality/available effects and of course disable EAX effects. (CPU usage is not important when writing the sounds to disk).