Oakrey.Audio 2.0.4

Oakrey.Audio

A Windows .NET library built on NAudio for audio recording, playback, synthesis, mixing, and signal analysis. The central type is FFTAudioSample, which wraps any ISampleProvider and adds an online STFT/FFT pipeline, reactive state tracking via IObservable<State>, and spectrum accumulation into SpectraList.

Main features

• FFT / STFT pipeline - FFTAudioSample feeds every read through SampleAggregator and accumulates frequency spectra in SpectraList. ShortTimeFourierTransformation provides a standalone STFT entry point.
• Playback - Player loads any audio file or ISampleProvider, supports configurable start/length offsets, volume control, output device selection, and SaveToFile.
• Synthesis - Synthesizers generates sine, square, sawtooth, pink noise, and sweep waveforms at configurable frequency and gain, ready to play or compare.
• Mixing - Mixer (playable) and SampleCreators.Mixer (analysis-only) wrap NAudio's MixingSampleProvider with automatic format conversion on AddAudioStream.
• Recording - FileRecorder streams microphone input directly to a WAV file. ObservableRecorder publishes a completed FFTAudioSample via IObservable<FFTAudioSample> on StopRecording.
• Audio comparators - ComparatorSNRSeg computes segmented SNR with Hann windowing and configurable overlap. ComparatorFwSNRSeg computes frequency-weighted segmented SNR. ComparatorLogLikelihoodRatio computes LLR using a Levinson-Durbin AR model.
• Device enumeration - AudioDevicesEnumerator lists all WaveIn / WaveOut devices by name and resolves device indices for recording and playback.
• File I/O - WaveAudioFileAccess loads a region of a WAV file into an FFTAudioSample and saves an FFTAudioSample back to a WAV file.

Architecture

classDiagram
    class FFTAudioSample {
        +SpectraList ListOfFrequencySpectra
        +IObservable~State~ Status
        +bool GetSpectra
        +Play() Pause() Stop() Resume() Record()
    }
    class AudioBasePlayable {
        +byte Volume
        +Play(deviceName)
        +SaveToFile(filename)
    }
    class RecorderBase {
        +TimeSpan RecordTime
        +bool SpectrumComputing
        +StartRecording()
        +StopRecording()
    }

    FFTAudioSample <|-- AudioBasePlayable
    FFTAudioSample <|-- RecorderBase
    AudioBasePlayable <|-- Player
    AudioBasePlayable <|-- Players_Mixer
    AudioBasePlayable <|-- Synthesizers
    RecorderBase <|-- FileRecorder
    RecorderBase <|-- ObservableRecorder

    class ComparatorBase
    ComparatorBase <|-- ComparatorSNRSeg
    ComparatorBase <|-- ComparatorFwSNRSeg
    ComparatorBase <|-- ComparatorLogLikelihoodRatio

Project structure

Requirements

• .NET 10 or higher
• Windows only (net10.0-windows, uses WaveInEvent / WaveOutEvent)
• NAudio 2.2.1

Installation

dotnet add package Oakrey.Audio

Or via the Visual Studio NuGet Package Manager: search for Oakrey.Audio.

Example usage

Play a WAV file on a specific output device

using Oakrey.Audio.Players;

using Player player = Player.Create("speech.wav");
player.PlaybackStart = TimeSpan.FromSeconds(2);
player.PlaybackLength = TimeSpan.FromSeconds(5);
player.Volume = 80;
player.Play("Speakers (Realtek High Definition Audio)");

Record microphone input to a file

using Oakrey.Audio.Players;

using FileRecorder recorder = FileRecorder.Create("capture.wav", "Microphone Array");
recorder.RecordTime = TimeSpan.FromSeconds(10);
recorder.StartRecording();
// ... wait ...
recorder.StopRecording();

Reactive recording with FFT spectrum

using Oakrey.Audio;
using Oakrey.Audio.Players;

using ObservableRecorder recorder = ObservableRecorder.Create("Microphone Array");
recorder.SpectrumComputing = true;

using IDisposable sub = recorder.SampleRecorded.Subscribe(sample =>
{
    SpectraList spectra = sample.ListOfFrequencySpectra;
    Console.WriteLine($"Captured {spectra.Count} spectra frames");
});

recorder.StartRecording();
await Task.Delay(TimeSpan.FromSeconds(3));
recorder.StopRecording();

Compare two audio samples with segmented SNR

using Oakrey.Audio;
using Oakrey.Audio.AudioComparators;
using Oakrey.Audio.SampleCreators;

FFTAudioSample reference = WaveAudioFileAccess.LoadFromFile("ref.wav", TimeSpan.Zero, TimeSpan.FromSeconds(5));
FFTAudioSample distorted = WaveAudioFileAccess.LoadFromFile("dist.wav", TimeSpan.Zero, TimeSpan.FromSeconds(5));

ComparatorSNRSeg comparator = new(reference, distorted);
float snr = comparator.Analyze(CancellationToken.None);
Console.WriteLine($"SNR: {snr:F2} dB");

Synthesize and play a sine wave

using Oakrey.Audio.Players;

using Synthesizers synth = Synthesizers.CreateSine(TimeSpan.FromSeconds(3), frequency: 440.0);
synth.Volume = 60;
synth.Play();

Development notes

• FFTAudioSample inherits from DisposableBase (Oakrey.Disposables). Always dispose instances created with new or factory methods.
• The FFT window length is fixed at 2048 samples. Spectrum frames are accumulated in SpectraList only while GetSpectra = true.
• RecorderBase.RecordTime = TimeSpan.Zero means unlimited recording duration.
• AudioDevicesEnumerator includes device index -1 (the Windows default device mapper) in its enumeration; index -1 is valid for WaveIn/WaveOut.
• ComparatorLogLikelihoodRatio uses a Levinson-Durbin AR model internally; AR order defaults apply and the computation supports CancellationToken.
• The library targets net10.0-windows exclusively; it cannot be used in cross-platform or server-side (Linux/macOS) builds.

License

MIT - Copyright (c) Oakrey 2016-present. Repository: https://dev.azure.com/oakrey/OpenPackages/_git/Drivers