Audio graphs

This article shows how to use the APIs in the Windows.Media.Audio namespace to create audio graphs for audio routing, mixing, and processing scenarios.

An audio graph is a set of interconnected audio nodes through which audio data flows.

  • Audio input nodes supply audio data to the graph from audio input devices, audio files, or from custom code. lat

  • Audio output nodes are the destination for audio processed by the graph. Audio can be routed out of the graph to audio output devices, audio files, or custom code.

  • Submix nodes take audio from one or more nodes and combine them into a single output that can be routed to other nodes in the graph.

After all of the nodes have been created and the connections between them set up, you simply start the audio graph and the audio data flows from the input nodes, through any submix nodes, to the output nodes. This model makes scenarios such as recording from a device's microphone to an audio file, playing audio from a file to a device's speaker, or mixing audio from multiple sources quick and easy to implement.

Additional scenarios are enabled with the addition of audio effects to the audio graph. Every node in an audio graph can be populated with zero or more audio effects that perform audio processing on the audio passing through the node. There are several built-in effects such as echo, equalizer, limiting, and reverb that can be attached to an audio node with just a few lines of code. You can also create your own custom audio effects that work exactly the same as the built-in effects.

Note

The AudioGraph UWP sample implements the code discussed in this overview. You can download the sample to see the code in context or to use as a starting point for your own app.

Choosing Windows Runtime AudioGraph or XAudio2

The Windows Runtime audio graph APIs offer functionality that can also be implemented by using the COM-based XAudio2 APIs. The following are features of the Windows Runtime audio graph framework that differ from XAudio2.

The Windows Runtime audio graph APIs:

  • Are significantly easier to use than XAudio2.
  • Can be used from C# in addition to being supported for C++.
  • Can use audio files, including compressed file formats, directly. XAudio2 only operates on audio buffers and does not provide any file I/O capabilities.
  • Can use the low-latency audio pipeline in Windows 10.
  • Support automatic endpoint switching when default endpoint parameters are used. For example, if the user switches from a device's speaker to a headset, the audio is automatically redirected to the new input.

AudioGraph class

The AudioGraph class is the parent of all nodes that make up the graph. Use this object to create instances of all of the audio node types. Create an instance of the AudioGraph class by initializing an AudioGraphSettings object containing configuration settings for the graph, and then calling AudioGraph.CreateAsync. The returned CreateAudioGraphResult gives access to the created audio graph or provides an error value if audio graph creation fails.

AudioGraph audioGraph;
private async Task InitAudioGraph()
{

    AudioGraphSettings settings = new AudioGraphSettings(Windows.Media.Render.AudioRenderCategory.Media);

    CreateAudioGraphResult result = await AudioGraph.CreateAsync(settings);
    if (result.Status != AudioGraphCreationStatus.Success)
    {
        ShowErrorMessage("AudioGraph creation error: " + result.Status.ToString());
    }

    audioGraph = result.Graph;

}
  • All audio node types are created by using the Create* methods of the AudioGraph class.

  • The AudioGraph.Start method causes the audio graph to start processing audio data. The AudioGraph.Stop method stops audio processing. Each node in the graph can be started and stopped independently while the graph is running, but no nodes are active when the graph is stopped. ResetAllNodes causes all nodes in the graph to discard any data currently in their audio buffers.

  • The QuantumStarted event occurs when the graph is starting the processing of a new quantum of audio data. The QuantumProcessed event occurs when the processing of a quantum is completed.

  • The only AudioGraphSettings property that is required is AudioRenderCategory. Specifying this value allows the system to optimize the audio pipeline for the specified category.

  • The quantum size of the audio graph determines the number of samples that are processed at one time. By default, the quantum size is 10 ms based at the default sample rate. If you specify a custom quantum size by setting the DesiredSamplesPerQuantum property, you must also set the QuantumSizeSelectionMode property to ClosestToDesired or the supplied value is ignored. If this value is used, the system will choose a quantum size as close as possible to the one you specify. To determine the actual quantum size, check the SamplesPerQuantum of the AudioGraph after it has been created.

  • If you only plan to use the audio graph with files and don't plan to output to an audio device, it is recommended that you use the default quantum size by not setting the DesiredSamplesPerQuantum property.

  • The DesiredRenderDeviceAudioProcessing property determines the amount of processing the primary render device performs on the output of the audio graph. The Default setting allows the system to use the default audio processing for the specified audio render category. This processing can significantly improve the sound of audio on some devices, particularly mobile devices with small speakers. The Raw setting can improve performance by minimizing the amount of signal processing performed, but can result in inferior sound quality on some devices.

  • If the QuantumSizeSelectionMode is set to LowestLatency, the audio graph will automatically use Raw for DesiredRenderDeviceAudioProcessing.

  • Starting with Windows 10, version 1803, you can set the AudioGraphSettings.MaxPlaybackSpeedFactor property to set a maximum value used for the AudioFileInputNode.PlaybackSpeedFactor, AudioFrameInputNode.PlaybackSpeedFactor, and MediaSourceInputNode.PlaybackSpeedFactor properties. When an audio graph supports a playback speed factor greater than 1, the system must allocate additional memory in order to maintain a sufficient buffer of audio data. For this reason, setting MaxPlaybackSpeedFactor to the lowest value required by your app will reduce the memory consumption of your app. If your app will only play back content at normal speed, it is recommended that you set MaxPlaybackSpeedFactor to 1.

  • The EncodingProperties determines the audio format used by the graph. Only 32-bit float formats are supported.

  • The PrimaryRenderDevice sets the primary render device for the audio graph. If you don't set this, the default system device is used. The primary render device is used to calculate the quantum sizes for other nodes in the graph. If there are no audio render devices present on the system, audio graph creation will fail.

You can let the audio graph use the default audio render device or use the Windows.Devices.Enumeration.DeviceInformation class to get a list of the system's available audio render devices by calling FindAllAsync and passing in the audio render device selector returned by Windows.Media.Devices.MediaDevice.GetAudioRenderSelector. You can choose one of the returned DeviceInformation objects programmatically or show UI to allow the user to select a device and then use it to set the PrimaryRenderDevice property.

Windows.Devices.Enumeration.DeviceInformationCollection devices =
 await Windows.Devices.Enumeration.DeviceInformation.FindAllAsync(Windows.Media.Devices.MediaDevice.GetAudioRenderSelector());

// Show UI to allow the user to select a device
Windows.Devices.Enumeration.DeviceInformation selectedDevice = ShowMyDeviceSelectionUI(devices);


settings.PrimaryRenderDevice = selectedDevice;

Device input node

A device input node feeds audio into the graph from an audio capture device connected to the system, such as a microphone. Create a DeviceInputNode object that uses the system's default audio capture device by calling CreateDeviceInputNodeAsync. Provide an AudioRenderCategory to allow the system to optimize the audio pipeline for the specified category.

AudioDeviceInputNode deviceInputNode;
private async Task CreateDeviceInputNode()
{
    // Create a device output node
    CreateAudioDeviceInputNodeResult result = await audioGraph.CreateDeviceInputNodeAsync(Windows.Media.Capture.MediaCategory.Media);

    if (result.Status != AudioDeviceNodeCreationStatus.Success)
    {
        // Cannot create device output node
        ShowErrorMessage(result.Status.ToString());
        return;
    }

    deviceInputNode = result.DeviceInputNode;
}

If you want to specify a specific audio capture device for the device input node, you can use the Windows.Devices.Enumeration.DeviceInformation class to get a list of the system's available audio capture devices by calling FindAllAsync and passing in the audio render device selector returned by Windows.Media.Devices.MediaDevice.GetAudioCaptureSelector. You can choose one of the returned DeviceInformation objects programmatically or show UI to allow the user to select a device and then pass it into CreateDeviceInputNodeAsync.

Windows.Devices.Enumeration.DeviceInformationCollection devices =
 await Windows.Devices.Enumeration.DeviceInformation.FindAllAsync(Windows.Media.Devices.MediaDevice.GetAudioCaptureSelector());

// Show UI to allow the user to select a device
Windows.Devices.Enumeration.DeviceInformation selectedDevice = ShowMyDeviceSelectionUI(devices);

CreateAudioDeviceInputNodeResult result =
    await audioGraph.CreateDeviceInputNodeAsync(Windows.Media.Capture.MediaCategory.Media, audioGraph.EncodingProperties, selectedDevice);

Device output node

A device output node pushes audio from the graph to an audio render device, such as speakers or a headset. Create a DeviceOutputNode by calling CreateDeviceOutputNodeAsync. The output node uses the PrimaryRenderDevice of the audio graph.

AudioDeviceOutputNode deviceOutputNode;
private async Task CreateDeviceOutputNode()
{
    // Create a device output node
    CreateAudioDeviceOutputNodeResult result = await audioGraph.CreateDeviceOutputNodeAsync();

    if (result.Status != AudioDeviceNodeCreationStatus.Success)
    {
        // Cannot create device output node
        ShowErrorMessage(result.Status.ToString());
        return;
    }

    deviceOutputNode = result.DeviceOutputNode;
}

File input node

A file input node allows you to feed data from an audio file into the graph. Create an AudioFileInputNode by calling CreateFileInputNodeAsync.

AudioFileInputNode fileInputNode;
private async Task CreateFileInputNode()
{
    if (audioGraph == null)
        return;

    FileOpenPicker filePicker = new FileOpenPicker();
    filePicker.SuggestedStartLocation = PickerLocationId.MusicLibrary;
    filePicker.FileTypeFilter.Add(".mp3");
    filePicker.FileTypeFilter.Add(".wav");
    filePicker.FileTypeFilter.Add(".wma");
    filePicker.FileTypeFilter.Add(".m4a");
    filePicker.ViewMode = PickerViewMode.Thumbnail;
    StorageFile file = await filePicker.PickSingleFileAsync();

    // File can be null if cancel is hit in the file picker
    if (file == null)
    {
        return;
    }
    CreateAudioFileInputNodeResult result = await audioGraph.CreateFileInputNodeAsync(file);

    if (result.Status != AudioFileNodeCreationStatus.Success)
    {
        ShowErrorMessage(result.Status.ToString());
    }

    fileInputNode = result.FileInputNode;
}
  • File input nodes support the following file formats: mp3, wav, wma, m4a.
  • Set the StartTime property to specify the time offset into the file where playback should begin. If this property is null, the beginning of the file is used. Set the EndTime property to specify the time offset into the file where playback should end. If this property is null, the end of the file is used. The start time value must be lower than the end time value, and the end time value must be less than or equal to the duration of the audio file, which can be determined by checking the Duration property value.
  • Seek to a position in the audio file by calling Seek and specifying the time offset into the file to which the playback position should be moved. The specified value must be within the StartTime and EndTime range. Get the current playback position of the node with the read-only Position property.
  • Enable looping of the audio file by setting the LoopCount property. When non-null, this value indicates the number of times the file will be played in after the initial playback. So, for example, setting LoopCount to 1 will cause the file to be played 2 times in total, and setting it to 5 will cause the file to be played 6 times in total. Setting LoopCount to null causes the file to be looped indefinitely. To stop looping, set the value to 0.
  • Adjust the speed at which the audio file is played back by setting the PlaybackSpeedFactor. A value of 1 indicates the original speed of the file, .5 is half-speed, and 2 is double speed.

MediaSource input node

The MediaSource class provides a common way to reference media from different sources and exposes a common model for accessing media data regardless of the underlying media format which could be a file on disk, a stream, or an adaptive streaming network source. A **MediaSourceAudioInputNode node lets you direct audio data from a MediaSource into the audio graph. Create a MediaSourceAudioInputNode by calling CreateMediaSourceAudioInputNodeAsync, passing in a MediaSource object representing the content you wish to play. A **CreateMediaSourceAudioInputNodeResult is returned which you can use to determine the status of the operation by checking the Status property. If the status is Success, you can get the created MediaSourceAudioInputNode by accessing the Node property. The following example shows the creation of a node from an AdaptiveMediaSource object representing content streaming over the network. For more information on working with MediaSource, see Media items, playlists, and tracks. For more information on streaming media content over the internet, see Adaptive streaming.

MediaSourceAudioInputNode mediaSourceInputNode;
private async Task CreateMediaSourceInputNode(System.Uri contentUri)
{
    if (audioGraph == null)
        return;

    var adaptiveMediaSourceResult = await AdaptiveMediaSource.CreateFromUriAsync(contentUri);
    if(adaptiveMediaSourceResult.Status != AdaptiveMediaSourceCreationStatus.Success)
    {
        Debug.WriteLine("Failed to create AdaptiveMediaSource");
        return;
    }

    var mediaSource = MediaSource.CreateFromAdaptiveMediaSource(adaptiveMediaSourceResult.MediaSource);
    CreateMediaSourceAudioInputNodeResult mediaSourceAudioInputNodeResult =
        await audioGraph.CreateMediaSourceAudioInputNodeAsync(mediaSource);

    if (mediaSourceAudioInputNodeResult.Status != MediaSourceAudioInputNodeCreationStatus.Success)
    {
        switch (mediaSourceAudioInputNodeResult.Status)
        {
            case MediaSourceAudioInputNodeCreationStatus.FormatNotSupported:
                Debug.WriteLine("The MediaSource uses an unsupported format");
                break;
            case MediaSourceAudioInputNodeCreationStatus.NetworkError:
                Debug.WriteLine("The MediaSource requires a network connection and a network-related error occurred");
                break;
            case MediaSourceAudioInputNodeCreationStatus.UnknownFailure:
            default:
                Debug.WriteLine("An unknown error occurred while opening the MediaSource");
                break;
        }
        return;
    }

    mediaSourceInputNode = mediaSourceAudioInputNodeResult.Node;
}

To receive a notification when playback has reached the end of the MediaSource content, register a handler for the MediaSourceCompleted event.

mediaSourceInputNode.MediaSourceCompleted += MediaSourceInputNode_MediaSourceCompleted;
private void MediaSourceInputNode_MediaSourceCompleted(MediaSourceAudioInputNode sender, object args)
{
    audioGraph.Stop();
}

While playing a file from diskis likely to always complete successfully, media streamed from a network source may fail during playback due to a change in network connection or other issues that are outside the control of the audio graph. If a MediaSource becomes unplayable during playback, the audio graph will raise the UnrecoverableErrorOccurred event. You can use the handler for this event to stop and dispose of the audio graph and then reinitialize your graph.

audioGraph.UnrecoverableErrorOccurred += AudioGraph_UnrecoverableErrorOccurred;
private void AudioGraph_UnrecoverableErrorOccurred(AudioGraph sender, AudioGraphUnrecoverableErrorOccurredEventArgs args)
{
    if (sender == audioGraph && args.Error != AudioGraphUnrecoverableError.None)
    {
        Debug.WriteLine("The audio graph encountered and unrecoverable error.");
        audioGraph.Stop();
        audioGraph.Dispose();
        InitAudioGraph();
    }
}

File output node

A file output node lets you direct audio data from the graph into an audio file. Create an AudioFileOutputNode by calling CreateFileOutputNodeAsync.

AudioFileOutputNode fileOutputNode;
private async Task CreateFileOutputNode()
{
    FileSavePicker saveFilePicker = new FileSavePicker();
    saveFilePicker.FileTypeChoices.Add("Pulse Code Modulation", new List<string>() { ".wav" });
    saveFilePicker.FileTypeChoices.Add("Windows Media Audio", new List<string>() { ".wma" });
    saveFilePicker.FileTypeChoices.Add("MPEG Audio Layer-3", new List<string>() { ".mp3" });
    saveFilePicker.SuggestedFileName = "New Audio Track";
    StorageFile file = await saveFilePicker.PickSaveFileAsync();

    // File can be null if cancel is hit in the file picker
    if (file == null)
    {
        return;
    }

    Windows.Media.MediaProperties.MediaEncodingProfile mediaEncodingProfile;
    switch (file.FileType.ToString().ToLowerInvariant())
    {
        case ".wma":
            mediaEncodingProfile = MediaEncodingProfile.CreateWma(AudioEncodingQuality.High);
            break;
        case ".mp3":
            mediaEncodingProfile = MediaEncodingProfile.CreateMp3(AudioEncodingQuality.High);
            break;
        case ".wav":
            mediaEncodingProfile = MediaEncodingProfile.CreateWav(AudioEncodingQuality.High);
            break;
        default:
            throw new ArgumentException();
    }


    // Operate node at the graph format, but save file at the specified format
    CreateAudioFileOutputNodeResult result = await audioGraph.CreateFileOutputNodeAsync(file, mediaEncodingProfile);

    if (result.Status != AudioFileNodeCreationStatus.Success)
    {
        // FileOutputNode creation failed
        ShowErrorMessage(result.Status.ToString());
        return;
    }

    fileOutputNode = result.FileOutputNode;
}

Audio frame input node

An audio frame input node allows you to push audio data that you generate in your own code into the audio graph. This enables scenarios like creating a custom software synthesizer. Create an AudioFrameInputNode by calling CreateFrameInputNode.

AudioFrameInputNode frameInputNode;
private void CreateFrameInputNode()
{
    // Create the FrameInputNode at the same format as the graph, except explicitly set mono.
    AudioEncodingProperties nodeEncodingProperties = audioGraph.EncodingProperties;
    nodeEncodingProperties.ChannelCount = 1;
    frameInputNode = audioGraph.CreateFrameInputNode(nodeEncodingProperties);

    // Initialize the Frame Input Node in the stopped state
    frameInputNode.Stop();

    // Hook up an event handler so we can start generating samples when needed
    // This event is triggered when the node is required to provide data
    frameInputNode.QuantumStarted += node_QuantumStarted;
}

The FrameInputNode.QuantumStarted event is raised when the audio graph is ready to begin processing the next quantum of audio data. You supply your custom generated audio data from within the handler to this event.

private void node_QuantumStarted(AudioFrameInputNode sender, FrameInputNodeQuantumStartedEventArgs args)
{
    // GenerateAudioData can provide PCM audio data by directly synthesizing it or reading from a file.
    // Need to know how many samples are required. In this case, the node is running at the same rate as the rest of the graph
    // For minimum latency, only provide the required amount of samples. Extra samples will introduce additional latency.
    uint numSamplesNeeded = (uint)args.RequiredSamples;

    if (numSamplesNeeded != 0)
    {
        AudioFrame audioData = GenerateAudioData(numSamplesNeeded);
        frameInputNode.AddFrame(audioData);
    }
}
  • The FrameInputNodeQuantumStartedEventArgs object passed into the QuantumStarted event handler exposes the RequiredSamples property that indicates how many samples the audio graph needs to fill up the quantum to be processed.
  • Call AudioFrameInputNode.AddFrame to pass an AudioFrame object filled with audio data into the graph.
  • A new set of APIs for using MediaFrameReader with audio data were introduced in Windows 10, version 1803. These APIs allow you to obtain AudioFrame objects from a media frame source, which can be passed into a FrameInputNode using the AddFrame method. For more information, see Process audio frames with MediaFrameReader.
  • An example implementation of the GenerateAudioData helper method is shown below.

To populate an AudioFrame with audio data, you must get access to the underlying memory buffer of the audio frame. To do this you must initialize the IMemoryBufferByteAccess COM interface by adding the following code within your namespace.

[ComImport]
[Guid("5B0D3235-4DBA-4D44-865E-8F1D0E4FD04D")]
[InterfaceType(ComInterfaceType.InterfaceIsIUnknown)]
unsafe interface IMemoryBufferByteAccess
{
    void GetBuffer(out byte* buffer, out uint capacity);
}

The following code shows an example implementation of a GenerateAudioData helper method that creates an AudioFrame and populates it with audio data.

private double audioWaveTheta = 0;

unsafe private AudioFrame GenerateAudioData(uint samples)
{
    // Buffer size is (number of samples) * (size of each sample)
    // We choose to generate single channel (mono) audio. For multi-channel, multiply by number of channels
    uint bufferSize = samples * sizeof(float);
    AudioFrame frame = new Windows.Media.AudioFrame(bufferSize);

    using (AudioBuffer buffer = frame.LockBuffer(AudioBufferAccessMode.Write))
    using (IMemoryBufferReference reference = buffer.CreateReference())
    {
        byte* dataInBytes;
        uint capacityInBytes;
        float* dataInFloat;

        // Get the buffer from the AudioFrame
        ((IMemoryBufferByteAccess)reference).GetBuffer(out dataInBytes, out capacityInBytes);

        // Cast to float since the data we are generating is float
        dataInFloat = (float*)dataInBytes;

        float freq = 1000; // choosing to generate frequency of 1kHz
        float amplitude = 0.3f;
        int sampleRate = (int)audioGraph.EncodingProperties.SampleRate;
        double sampleIncrement = (freq * (Math.PI * 2)) / sampleRate;

        // Generate a 1kHz sine wave and populate the values in the memory buffer
        for (int i = 0; i < samples; i++)
        {
            double sinValue = amplitude * Math.Sin(audioWaveTheta);
            dataInFloat[i] = (float)sinValue;
            audioWaveTheta += sampleIncrement;
        }
    }

    return frame;
}
  • Because this method accesses the raw buffer underlying the Windows Runtime types, it must be declared using the unsafe keyword. You must also configure your project in Microsoft Visual Studio to allow the compilation of unsafe code by opening the project's Properties page, clicking the Build property page, and selecting the Allow Unsafe Code checkbox.
  • Initialize a new instance of AudioFrame, in the Windows.Media namespace, by passing in the desired buffer size to the constructor. The buffer size is the number of samples multiplied by the size of each sample.
  • Get the AudioBuffer of the audio frame by calling LockBuffer.
  • Get an instance of the IMemoryBufferByteAccess COM interface from the audio buffer by calling CreateReference.
  • Get a pointer to raw audio buffer data by calling IMemoryBufferByteAccess.GetBuffer and cast it to the sample data type of the audio data.
  • Fill the buffer with data and return the AudioFrame for submission into the audio graph.

Audio frame output node

An audio frame output node allows you to receive and process audio data output from the audio graph with custom code that you create. An example scenario for this is performing signal analysis on the audio output. Create an AudioFrameOutputNode by calling CreateFrameOutputNode.

AudioFrameOutputNode frameOutputNode;
private void CreateFrameOutputNode()
{
    frameOutputNode = audioGraph.CreateFrameOutputNode();
    audioGraph.QuantumStarted += AudioGraph_QuantumStarted;
}

The AudioGraph.QuantumStarted event is raised when the audio graph has begins processing a quantum of audio data. You can access the audio data from within the handler for this event.

Note

If you want to retrieve audio frames on a regular cadence, synchronized with the audio graph, call AudioFrameOutputNode.GetFrame from within the synchronous QuantumStarted event handler. The QuantumProcessed event is raised asynchronously after the audio engine has completed audio processing, which means its cadence may be irregular. Therefore you should not use the QuantumProcessed event for synchronized processing of audio frame data.

private void AudioGraph_QuantumStarted(AudioGraph sender, object args)
{
    AudioFrame frame = frameOutputNode.GetFrame();
    ProcessFrameOutput(frame);

}
  • Call GetFrame to get an AudioFrame object filled with audio data from the graph.
  • An example implementation of the ProcessFrameOutput helper method is shown below.
unsafe private void ProcessFrameOutput(AudioFrame frame)
{
    using (AudioBuffer buffer = frame.LockBuffer(AudioBufferAccessMode.Write))
    using (IMemoryBufferReference reference = buffer.CreateReference())
    {
        byte* dataInBytes;
        uint capacityInBytes;
        float* dataInFloat;

        // Get the buffer from the AudioFrame
        ((IMemoryBufferByteAccess)reference).GetBuffer(out dataInBytes, out capacityInBytes);

        dataInFloat = (float*)dataInBytes;
    }
}
  • Like the audio frame input node example above, you will need to declare the IMemoryBufferByteAccess COM interface and configure your project to allow unsafe code in order to access the underlying audio buffer.
  • Get the AudioBuffer of the audio frame by calling LockBuffer.
  • Get an instance of the IMemoryBufferByteAccess COM interface from the audio buffer by calling CreateReference.
  • Get a pointer to raw audio buffer data by calling IMemoryBufferByteAccess.GetBuffer and cast it to the sample data type of the audio data.

Node connections and submix nodes

All input nodes types expose the AddOutgoingConnection method that routes the audio produced by the node to the node that is passed into the method. The following example connects an AudioFileInputNode to an AudioDeviceOutputNode, which is a simple setup for playing an audio file on the device's speaker.

fileInputNode.AddOutgoingConnection(deviceOutputNode);

You can create more than one connection from an input node to other nodes. The following example adds another connection from the AudioFileInputNode to an AudioFileOutputNode. Now, the audio from the audio file is played to the device's speaker and is also written out to an audio file.

fileInputNode.AddOutgoingConnection(fileOutputNode);

Output nodes can also receive more than one connection from other nodes. In the following example a connection is made from a AudioDeviceInputNode to the AudioDeviceOutput node. Because the output node has connections from the file input node and the device input node, the output will contain a mix of audio from both sources. AddOutgoingConnection provides an overload that lets you specify a gain value for the signal passing through the connection.

deviceInputNode.AddOutgoingConnection(deviceOutputNode, .5);

Although output nodes can accept connections from multiple nodes, you may want to create an intermediate mix of signals from one or more nodes before passing the mix to an output. For example, you may want to set the level or apply effects to a subset of the audio signals in a graph. To do this, use the AudioSubmixNode. You can connect to a submix node from one or more input nodes or other submix nodes. In the following example, a new submix node is created with AudioGraph.CreateSubmixNode. Then, connections are added from a file input node and a frame output node to the submix node. Finally, the submix node is connected to a file output node.

private void CreateSubmixNode()
{
    AudioSubmixNode submixNode = audioGraph.CreateSubmixNode();
    fileInputNode.AddOutgoingConnection(submixNode);
    frameInputNode.AddOutgoingConnection(submixNode);
    submixNode.AddOutgoingConnection(fileOutputNode);
}

Starting and stopping audio graph nodes

When AudioGraph.Start is called, the audio graph begins processing audio data. Every node type provides Start and Stop methods that cause the individual node to start or stop processing data. When AudioGraph.Stop is called, all audio processing in the all nodes is stopped regardless of the state of individual nodes, but the state of each node can be set while the audio graph is stopped. For example, you could call Stop on an individual node while the graph is stopped and then call AudioGraph.Start, and the individual node will remain in the stopped state.

All node types expose the ConsumeInput property that, when set to false, allows the node to continue audio processing but stops it from consuming any audio data being input from other nodes.

All node types expose the Reset method that causes the node to discard any audio data currently in its buffer.

Adding audio effects

The audio graph API allows you to add audio effects to every type of node in a graph. Output nodes, input nodes, and submix nodes can each have an unlimited number of audio effects, limited only by the capabilities of the hardware.The following example demonstrates adding the built-in echo effect to a submix node.

EchoEffectDefinition echoEffect = new EchoEffectDefinition(audioGraph);
echoEffect.Delay = 1000.0;
echoEffect.Feedback = .2;
echoEffect.WetDryMix = .5;

submixNode.EffectDefinitions.Add(echoEffect);
  • All audio effects implement IAudioEffectDefinition. Every node exposes an EffectDefinitions property representing the list of effects applied to that node. Add an effect by adding it's definition object to the list.
  • There are several effect definition classes that are provided in the Windows.Media.Audio namespace. These include:
  • You can create your own audio effects that implement IAudioEffectDefinition and apply them to any node in an audio graph.
  • Every node type exposes a DisableEffectsByDefinition method that disables all effects in the node's EffectDefinitions list that were added using the specified definition. EnableEffectsByDefinition enables the effects with the specified definition.

Spatial audio

Starting with Windows 10, version 1607, AudioGraph supports spatial audio, which allows you to specify the location in 3D space from which audio from any input or submix node is emitted. You can also specify a shape and direction in which audio is emitted,a velocity that will be used to Doppler shift the node's audio, and define a decay model that describes how the audio is attenuated with distance.

To create an emitter, you can first create a shape in which the sound is projected from the emitter, which can be a cone or omnidirectional. The AudioNodeEmitterShape class provides static methods for creating each of these shapes. Next, create a decay model. This defines how the volume of the audio from the emitter decreases as the distance from the listener increases. The CreateNatural method creates a decay model that emulates the natural decay of sound using a distance squared falloff model. Finally, create an AudioNodeEmitterSettings object. Currently, this object is only used to enable and disable velocity-based Doppler attenuation of the emitter's audio. Call the AudioNodeEmitter constructor, passing in the initialization objects you just created. By default, the emitter is placed at the origin, but you can set the position of the emitter with the Position property.

Note

Audio node emitters can only process audio that is formatted in mono with a sample rate of 48kHz. Attempting to use stereo audio or audio with a different sample rate will result in an exception.

You assign the emitter to an audio node when you create it by using the overloaded creation method for the type of node you want. In this example, CreateFileInputNodeAsync is used to create a file input node from a specified file and the AudioNodeEmitter object you want to associate with the node.

var emitterShape = AudioNodeEmitterShape.CreateOmnidirectional();
var decayModel = AudioNodeEmitterDecayModel.CreateNatural(.1, 1, 10, 100);
var settings = AudioNodeEmitterSettings.None;

var emitter = new AudioNodeEmitter(emitterShape, decayModel, settings);
emitter.Position = new System.Numerics.Vector3(10, 0, 5);

CreateAudioFileInputNodeResult result = await audioGraph.CreateFileInputNodeAsync(file, emitter);

if (result.Status != AudioFileNodeCreationStatus.Success)
{
    ShowErrorMessage(result.Status.ToString());
}

fileInputNode = result.FileInputNode;

The AudioDeviceOutputNode that outputs audio from the graph to the user has a listener object, accessed with the Listener property, which represents the location, orientation, and velocity of the user in the 3D space. The positions of all of the emitters in the graph are relative to the position and orientation of the listener object. By default, the listener is located at the origin (0,0,0) facing forward along the Z axis, but you can set it's position and orientation with the Position and Orientation properties.

deviceOutputNode.Listener.Position = new System.Numerics.Vector3(100, 0, 0);
deviceOutputNode.Listener.Orientation = System.Numerics.Quaternion.CreateFromYawPitchRoll(0, (float)Math.PI, 0);

You can update the location, velocity, and direction of emitters at runtime to simulate the movement of an audio source through 3D space.

var emitter = fileInputNode.Emitter;
emitter.Position = newObjectPosition;
emitter.DopplerVelocity = newObjectPosition - oldObjectPosition;

You can also update the location, velocity, and orientation of the listener object at runtime to simulate the movement of the user through 3D space.

deviceOutputNode.Listener.Position = newUserPosition;

By default, spatial audio is calculated using Microsoft's head-relative transfer function (HRTF) algorithm to attenuate the audio based on its shape, velocity, and position relative to the listener. You can set the SpatialAudioModel property to FoldDown to use a simple stereo mix method of simulating spatial audio that is less accurate but requires less CPU and memory resources.

See also