AI Features

AI-powered audio processing. Available in OwnAudioSharp only — not in the Basic package.

⚠️

These features require the full OwnAudioSharp NuGet package. They are not available in OwnAudioSharp.Basic.
ONNX model files are not bundled in the NuGet package. They are downloaded automatically on first use via VocalRemoverModelManager.DownloadModelAsync() and stored in the user's local application data folder. No manual setup is required.

Vocal Remover

Separates vocals from instrumental using ONNX neural network models. Namespace: OwnaudioNET.Features.Vocalremover

C# — Download model & separate

using OwnaudioNET.Features.Vocalremover;

// Download the model on first run (skipped automatically if already present)
if (!VocalRemoverModelManager.IsModelAvailable(InternalModel.Best))
{
    await VocalRemoverModelManager.DownloadModelAsync(
        InternalModel.Best,
        new Progress<ModelDownloadProgress>(p =>
            Console.Write($"\rDownloading: {p.Percentage:F1}%")));
}

var options = new SimpleSeparationOptions
{
    Model                = InternalModel.Best,  // see model table below
    OutputDirectory      = "output/",
    DisableNoiseReduction = false               // true = skip noise reduction
};

using var separator = new SimpleAudioSeparationService(options);
separator.Initialize();

var result = separator.Separate("song.mp3");

Console.WriteLine(result.VocalsPath);       // path to vocals-only file
Console.WriteLine(result.InstrumentalPath); // path to instrumental-only file

VocalRemoverModelManager

Member	Description
`DefaultModelsDirectory`	Get/set the local folder where model files are stored. Defaults to `LocalApplicationData/OwnAudio/models/`. Override at startup to redirect storage.
`IsModelAvailable(model)`	Returns `true` if the model file is already present on disk.
`DownloadModelAsync(model, progress, cancel)`	Downloads a single model from HuggingFace. Skips silently if the file already exists. Reports `ModelDownloadProgress` with byte count and percentage.
`EnsureModelsAvailableAsync(models, progress, cancel)`	Downloads all models in the collection that are not yet present. Models are fetched sequentially.

Available Models

Model	Size	Speed	Quality	Best for
`InternalModel.Default`	~28 MB	Medium	Very Good	General purpose
`InternalModel.Best`	~50 MB	Slow	Excellent	Final quality separation
`InternalModel.Karaoke`	~28 MB	Medium	Excellent	Karaoke tracks, preserves vocals
`InternalModel.HTDemucs`	~166 MB	Slowest	Highest	Multi-stem (vocals, bass, drums, other)

SimpleSeparationOptions

Property	Type	Description
`Model`	`InternalModel`	Which ONNX model to use for separation.
`ModelPath`	`string?`	Optional path to a custom ONNX file. Takes precedence over `Model` when set and the file exists.
`OutputDirectory`	`string`	Directory where output files are written.
`DisableNoiseReduction`	`bool`	Set to `true` to skip post-processing noise reduction (default: `false`).
`EnableGPU`	`bool`	Use GPU acceleration via CoreML (macOS) or CUDA (Windows/Linux). Default: `true`.
`ChunkSizeSeconds`	`int`	Audio chunk size in seconds. `0` processes the entire file at once. Default: `15`.

Chord Detection

Detects musical chords, key, and tempo from audio files. Namespace: OwnaudioNET.Features

C# — From a file

var (chords, key, bpm) = ChordDetect.DetectFromFile("song.mp3", intervalSecond: 1.0f);

foreach (var chord in chords)
    Console.WriteLine($"{chord.TimeSeconds:F1}s  {chord.ChordName}");

Console.WriteLine($"Key: {key}  BPM: {bpm}");

C# — From multiple files (mixed)

// Mixes all files before analysis
var (chords, key, bpm) = ChordDetect.DetectFromFiles(
    new[] { "guitar.wav", "piano.wav" },
    intervalSecond: 0.5f);

C# — Real-time detection

// Continuous detection from an active note stream
var (chord, stability) = ChordDetect.DetectRealtime(
    notes,                   // List<Note>
    DetectionMode.Standard,
    bufferSize: 4096);

Console.WriteLine($"Chord: {chord}  Stability: {stability:P0}");

Return Types

Member	Type	Description
`chords`	`List<TimedChord>`	List of chords with timestamps.
`TimedChord.TimeSeconds`	`double`	Timestamp in seconds.
`TimedChord.ChordName`	`string`	Chord name (e.g. `"Cmaj"`, `"Am7"`).
`key`	`MusicalKey`	Detected musical key.
`bpm`	`int`	Detected tempo in BPM.
`stability`	`float`	Real-time detection confidence (0.0–1.0).

Auto-Mastering (Matchering)

AI-driven EQ matching and automated audio mastering. Namespace: OwnaudioNET.Features

C# — Analyze spectrum

AudioSpectrum spectrum = AudioAnalyzer.AnalyzeAudioFile("track.wav");

C# — EQ match to reference

AudioAnalyzer.ProcessEQMatching(
    sourceFile: "my-mix.wav",
    targetFile: "reference.wav",   // the "sound" to match
    outputFile: "mastered.wav");

C# — Apply playback system preset

AudioAnalyzer.ProcessWithEnhancedPreset(
    sourceFile: "my-mix.wav",
    outputFile: "for-club.wav",
    system:     PlaybackSystem.Club,  // Club, Car, Studio, HomeTheater
    eqOnlyMode: true);

C# — Batch process

AudioAnalyzer.BatchProcessWithEnhancedPreset(
    sourceFiles:    new[] { "track1.wav", "track2.wav", "track3.wav" },
    baseSampleFile: "reference.wav",
    outputDirectory: "output/",
    system:         PlaybackSystem.Studio,
    fileNameSuffix: "_mastered");

PlaybackSystem Values

Value	Description
`PlaybackSystem.Club`	Optimized for nightclub / dance floor sound systems
`PlaybackSystem.Car`	Optimized for in-car audio systems
`PlaybackSystem.Studio`	Neutral studio monitor reference
`PlaybackSystem.HomeTheater`	Home theater speaker systems