XAudio2.Voice.FilterParameters Access Violation Issue

[as00374]

I am having some issues with XAudio2. I am attempting to implement single-voice-multiple-output using SourceVoice.SetOutputVoices by creating an array of VoiceSendDescriptors, but I get access violations for the filter parameters for all voices of any type, making this process fail. Can anyone tell me why this is and how I can fix it (Code below). Any help appreciates. Thanks:

using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Threading;
using System.IO;
using System.Threading.Tasks;
using SharpDX;
using SharpDX.XAudio2;
using SharpDX.Multimedia;

namespace ConsoleApplication4
{
    class Program : ThreadStaticAttribute
    {
        public const float pan = -1;
        /// <summary>
        /// SharpDX XAudio2 sample. Plays wav/xwma/adpcm files from the disk.
        /// </summary>
        static void Main(string[] args)
        {
            XAudio2 search = new XAudio2();
            XAudio2[] xaudio2 = new XAudio2[3];
            MasteringVoice[] masteringVoice = new MasteringVoice[3];
            int outputsFound = 0;

            for (int i = 0; i < search.DeviceCount; i++)
            {
                if (search.GetDeviceDetails(i).DisplayName.Contains("1-2"))
                {
                    xaudio2[0] = new XAudio2();
                    masteringVoice[0] = new MasteringVoice(xaudio2[0], 2, 48000, i);
                    outputsFound++;
                }
                else if (search.GetDeviceDetails(i).DisplayName.Contains("3-4"))
                {
                    xaudio2[1] = new XAudio2();
                    masteringVoice[1] = new MasteringVoice(xaudio2[1], 2, 48000, i);
                    outputsFound++;
                }
                else if (search.GetDeviceDetails(i).DisplayName.Contains("5-6"))
                {
                    xaudio2[2] = new XAudio2();
                    masteringVoice[2] = new MasteringVoice(xaudio2[2], 2, 48000, i);
                    outputsFound++;
                }
            }
            if (outputsFound == 3)
            {
                // No pan
                PlaySoundFile(xaudio2, masteringVoice, "1) Playing a standard WAV file- no pan", "ergon.wav", 0);

                // Pan
                PlaySoundFile(xaudio2, masteringVoice, "2) Playing a XWMA file- Device 1 panned left, Device 2 panned right, Device 3 no panning", "ergon.xwma", 1);

                // Reverse pan
                PlaySoundFile(xaudio2, masteringVoice, "3) Playing an ADPCM file- Device 1 panned right, Device 2 panned left, Device 3 no panning", "ergon.adpcm.wav", 2);

                for (int i = 0; i < xaudio2.Count(); i++)
                {
                    masteringVoice[i].Dispose();
                    xaudio2[i].Dispose();
                }
            }
            else
            {
                Console.WriteLine("Devices missing. Check UA 1-2, UA 3-4 and UA 5-6");
                throw new Exception("Devices missing. Check UA 1-2, UA 3-4 and UA 5-6");
            }
        }

        /// <summary>
        /// Play a sound file. Supported format are Wav(pcm+adpcm) and XWMA
        /// </summary>
        /// <param name="device">The device.</param>
        /// <param name="text">Text to display</param>
        /// <param name="fileName">Name of the file.</param>
        [STAThread]
        static void PlaySoundFile(XAudio2[] device, MasteringVoice[] mVoices, string text, string fileName, int panning)
        {
            Console.WriteLine("{0} => {1} (Press esc to skip)", text, fileName);
            var stream = new SoundStream(File.OpenRead(fileName));
            var waveFormat = stream.Format;
            var buffer = new AudioBuffer
            {
                Stream = stream.ToDataStream(),
                AudioBytes = (int)stream.Length,
                Flags = BufferFlags.EndOfStream
            };
            stream.Close();
            if(device.Count() >= 3) {
                var sourceVoice = new SourceVoice(device[0], waveFormat, true);
                VoiceSendDescriptor[] sendDescriptors = new VoiceSendDescriptor[3];
                for (int i = 0; i < sendDescriptors.Count(); i++)
                {
                    sendDescriptors[i] = new VoiceSendDescriptor(mVoices[i]);
                }
                sourceVoice.SetOutputVoices(sendDescriptors);
                // Adds a sample callback to check that they are working on source voices
                if (panning != 0)
                {
                    DeviceDetails deviceDetails = device[0].GetDeviceDetails(0);
                    VoiceDetails voiceDetails = sourceVoice.VoiceDetails;
                    float[][] outputMatrix = new float[device.Count()][];
                    outputMatrix[0] = new float[deviceDetails.OutputFormat.Channels * voiceDetails.InputChannelCount];
                    outputMatrix[1] = new float[deviceDetails.OutputFormat.Channels * voiceDetails.InputChannelCount];
                    outputMatrix[2] = new float[deviceDetails.OutputFormat.Channels * voiceDetails.InputChannelCount];

                    for (int i = 0; i < device.Count(); i++)
                    {
                        for (int j = 0; j < deviceDetails.OutputFormat.Channels * voiceDetails.InputChannelCount; j++)
                            outputMatrix[i][j] = 0;
                    }
                    if (panning == 1)
                    {
                        outputMatrix[0][0] = 0.5f - pan / 2;
                        outputMatrix[0][1] = 0.5f - pan / 2;
                        outputMatrix[0][2] = 0.5f + pan / 2;
                        outputMatrix[0][3] = 0.5f + pan / 2;

                        outputMatrix[1][0] = 0.5f + pan / 2;
                        outputMatrix[1][1] = 0.5f + pan / 2;
                        outputMatrix[1][2] = 0.5f - pan / 2;
                        outputMatrix[1][3] = 0.5f - pan / 2;

                        outputMatrix[1][0] = 1;
                        outputMatrix[1][1] = 0;
                        outputMatrix[1][2] = 0;
                        outputMatrix[1][3] = 1;
                    }
                    else
                    {
                        outputMatrix[1][0] = 0.5f - pan / 2;
                        outputMatrix[1][1] = 0.5f - pan / 2;
                        outputMatrix[1][2] = 0.5f + pan / 2;
                        outputMatrix[1][3] = 0.5f + pan / 2;

                        outputMatrix[0][0] = 0.5f + pan / 2;
                        outputMatrix[0][1] = 0.5f + pan / 2;
                        outputMatrix[0][2] = 0.5f - pan / 2;
                        outputMatrix[0][3] = 0.5f - pan / 2;

                        outputMatrix[1][0] = 1;
                        outputMatrix[1][1] = 0;
                        outputMatrix[1][2] = 0;
                        outputMatrix[1][3] = 1;
                    }
                    sourceVoice.SetOutputMatrix(sendDescriptors[0].OutputVoice, 2, 2, outputMatrix[0]);
                    sourceVoice.SetOutputMatrix(sendDescriptors[1].OutputVoice, 2, 2, outputMatrix[1]);
                    sourceVoice.SetOutputMatrix(sendDescriptors[2].OutputVoice, 2, 2, outputMatrix[2]);

                }
                sourceVoice.BufferEnd += (context) => Console.WriteLine(" => event received: end of buffer");
                sourceVoice.SubmitSourceBuffer(buffer, stream.DecodedPacketsInfo);
                sourceVoice.Start();

                int count = 0;
                while (sourceVoice.State.BuffersQueued > 0 && !IsKeyPressed(ConsoleKey.Escape))
                {
                    if (count == 50)
                    {
                        Console.Write(".");
                        Console.Out.Flush();
                        count = 0;
                    }
                    Thread.Sleep(10);
                    count++;
                }
                Console.WriteLine();

                sourceVoice.DestroyVoice();
                sourceVoice.Dispose();
                buffer.Stream.Dispose();
            }
        }

        /// <summary>
        /// Determines whether the specified key is pressed.
        /// </summary>
        /// <param name="key">The key.</param>
        /// <returns>
        ///   <c>true</c> if the specified key is pressed; otherwise, <c>false</c>.
        /// </returns>
        public static bool IsKeyPressed(ConsoleKey key)
        {
            return Console.KeyAvailable && Console.ReadKey(true).Key == key;
        }
    }
}

[ArtiomCiumac]

This is more likely an XAudio2 usage issue, I am not sure if you will get much help here. Try to enable native debugging. If you are on Windows 7 - try to enable XAudio debugging.

Btw, why you put STAThread to called methods instead of Main? This looks quite strange. Why do you need the base class ThreadStaticAttribute there?

[ArtiomCiumac]

@as00374, have you made any progress on this issue?

[as00374]

Thanks for looking into this @ArtiomCiumac. I have changed from using SharpDX to NAudio as native ASIO support was the easiest solution to this issue. FTW, I was using STAThread because audio process is real-time critical, therefore I don't want multiple threads accessing it. I wasn't running the code on the main because my main thread is Kinect skeletal tracking using real-time object tracking and didn't want to be doing image processing and sensitive audio processing on the same thread.

Thanks again

[ArtiomCiumac]

I still don't understand if there is a problem with SharpDX itself or its usage. Try to start with a sample and modify it to suit your needs. Also, I am not sure if you need the STA attribute - as this is more related to UI like WinForms or WPF, but DirectX works fine in multithreaded environments.

If you have time - have a look at the SharpDX.Toolkit.Audio - it provides a simplified wrapper over the XAudio2 API.

I will close this issue for now - feel free to reopen it if you you believe your usage is correct and there is an issue with SharpDX.

[mattAtCSL]

I deleted my post after I saw there was a different way of setting the effects, which I've now got working. I'm not sure if the other way should work, I was looking at C++ XAudio2 examples and trying to do the same.

I'll stick my test code back up if needed.