Streaming directly from mic to speakers using DirectSound

Greetings again,

I've continued to play with this some. The following is my current version. I've tried increasing the number of notification events, I've tried using a memory stream in between the capture buffer read and the secondary buffer write. I've tried using a CopyTo to a buffer[] from the capture buffer before doing the secondary buffer write using the byte[] buffer. In this version, I'm just doing a simple assignment to a byte[] buffer and then using that as a parameter to the secondary buffer write.

The all compile and build fine. They just don't produce any sound. I must be missing something really basic. Still hoping someone can provide me some assistance, perhaps someone from Microsoft?

Thanks,
FM

namespace Echo
{
public partial class Echo : Form
{
private Thread _echoThread;
private Capture _captureDevice;
private CaptureBuffer _captureBuffer;
private Device _playbackDevice;
private SecondaryBuffer _playbackBuffer;
private int _bufferSize;
private const int _bufferPositions = 10;
private AutoResetEvent _notificationEvent = null;
private BufferPositionNotify[] _positionNotify = new BufferPositionNotify[_bufferPositions + 1];
private Notify _echoNotify = null;
private void EchoThread()
{
_captureBuffer.Start(true);

int offset = 0;
for (; ; ) {
_notificationEvent.WaitOne(Timeout.Infinite, true);

byte[] buffer = (byte[]) _captureBuffer.Read(offset, typeof(byte), LockFlag.None, _bufferSize);
_playbackBuffer.Write(offset, buffer, LockFlag.None);
offset = (offset + _bufferSize) % (_bufferPositions * _bufferSize);

if (offset == 0)
console.AppendText(".");
}
}

public Echo()
{
CheckForIllegalCrossThreadCalls = false;
InitializeComponent();

console.AppendText("Echo\r\n");
console.AppendText("Written by Frank W. Miller\r\n");
console.AppendText("\r\n");

_captureDevice = new Capture();

short channels = 2;
console.AppendText("Channels: " + channels.ToString() + "\r\n");

short bitsPerSample = 16;
console.AppendText("Bits per sample: " + bitsPerSample.ToString() + "\r\n");

int samplesPerSecond = 22050;
console.AppendText("Samples per second: " + samplesPerSecond.ToString() + "\r\n");

//Set up the wave format to be captured
WaveFormat waveFormat = new WaveFormat();
waveFormat.Channels = channels;
waveFormat.FormatTag = WaveFormatTag.Pcm;
waveFormat.SamplesPerSecond = samplesPerSecond;
waveFormat.BitsPerSample = bitsPerSample;
waveFormat.BlockAlign = (short) (channels * (bitsPerSample / (short) 8));
waveFormat.AverageBytesPerSecond = waveFormat.BlockAlign * samplesPerSecond;

_bufferSize = waveFormat.AverageBytesPerSecond / 5;
console.AppendText("Buffer size: " + _bufferSize.ToString() + "\r\n");

CaptureBufferDescription captureBufferDescription = new CaptureBufferDescription();
captureBufferDescription.BufferBytes = _bufferPositions * _bufferSize;
captureBufferDescription.Format = waveFormat;
_captureBuffer = new CaptureBuffer(captureBufferDescription, _captureDevice);

_playbackDevice = new Device();
_playbackDevice.SetCooperativeLevel(this, CooperativeLevel.Normal);

BufferDescription playbackBufferDescription = new BufferDescription();
playbackBufferDescription.BufferBytes = _bufferPositions * _bufferSize;
playbackBufferDescription.Format = waveFormat;
_playbackBuffer = new SecondaryBuffer(playbackBufferDescription, _playbackDevice);

_echoThread = new Thread(new ThreadStart(EchoThread));
_echoThread.Start();

_notificationEvent = new AutoResetEvent(false);
for (int i = 0; i < _bufferPositions; i++) {
_positionNotify.Offset = (_bufferSize * i) + _bufferSize - 1;
_positionNotify.EventNotifyHandle = _notificationEvent.Handle;
}
_echoNotify = new Notify(_captureBuffer);
_echoNotify.SetNotificationPositions(_positionNotify, _bufferPositions);
}

private void EchoClose(object sender, FormClosingEventArgs e)
{
_echoThread.Abort();
}
}

FrankMiller at 2007-10-8 >

I really think there's something basic that I'm missing. For example, I'm wondering if the capture buffer is actually pointing into the output buffer of the microphone hardware and whether the secondary buffer points into the actual hardware output buffer for the speakers. If thats the case, then you would need to copy between them. Thats why I tried using CopyTo() to a byte[] buffer between the read and write. But that didnt work. But I may not have done it right.

I really wish somebody from Microsoft would comment on this. It would even be helpful to hear that it just can't be done. That way I could stop banging on it and do what other people do, which is to wrap native interfaces using interop. I want to avoid this mainly because I want to take this code to WM5 at some point and the DLLs that I could use under XP probably won't be there under WM5.

Thanks,
FM

FrankMiller at 2007-10-8 >