repositories {
    mavenCentral()
}

dependencies {
    implementation("io.github.sor2171:kaudioanalyser:1.4.2")
}

println(getNoteData(512, 440, NoteNameStyle.Scientific))
// NoteData(name=C5, cent=-37.63199, frequency=512.0, a4=440.0, style=Scientific)

val frequency1 = getFrequency("D5", 440f) // 587.3295f
val frequency2 = "E3".getNoteFrequency() // 164.81378f
val frequency3 = "E3".getNoteFrequency(432) // 161.81717f

println(NoteData(name = "B6", a4 = 440f).fillAll())
// NoteData(name=B6, cent=0.0, frequency=1975.5334, a4=440.0, style=Scientific)

println(NoteData(frequency = 1024f, a4 = 440f).fillAll())
// NoteData(name=C6, cent=-37.63199, frequency=1024.0, a4=440.0, style=Scientific)

println(NoteData(frequency = 1024f, a4 = 440f, style = NoteNameStyle.Helmholtz).fillAll())
// NoteData(name=c''', cent=-37.63199, frequency=1024.0, a4=440.0, style=Helmholtz)

// if used in KMP with Kodio
var audioQuality by rememberSaveable { mutableStateOf(AudioQuality.Standard) }
var bufferSize by rememberSaveable { mutableStateOf(2048) }
// for YIN algorithm, this buffer size could be enough
var detectedFrequency by remember { mutableStateOf(0f) }

val detector = remember(audioQuality, bufferSize) {
    PitchDetector(audioQuality.format.sampleRate, bufferSize)
}
LaunchedEffect(
    isRecording,
    isPausing,
    audioQuality,
    bufferSize
) {
    if (isRecording && !isPausing) {
        withContext(Dispatchers.Default) {
            val recorder = Kodio.recorder(quality = audioQuality)
            try {
                delay(500) // wait for the recorder to be released
                recorder.start()
                recorder.liveAudioFlow?.toAudioWindows(
                    windowSize = bufferSize,
                    hopSize = bufferSize / 2,
                    channels = audioQuality.format.channels.count,
                    bytesPerSample = audioQuality.format.bytesPerSample
                )?.collect { floatWindow ->

                    val frequency = detector.detectPitchYIN(floatWindow)
                    // or you can use FFT-based pitch detection, like this:
                    // val frequency = detector.detectPitch(floatWindow)
                    // it needs larger buffer size to be accurate

                    withContext(Dispatchers.Main) {
                        // stable the frequency by averaging with previous value
                        detectedFrequency = if (frequency > 0f) {
                            if (detectedFrequency == 0f) frequency
                            else detectedFrequency * 0.5f + frequency * 0.5f
                        } else {
                            0f
                        }
                    }
                }
            } finally {
                recorder.release()
            }
        }
    }
}

// also, used in KMP with Kodio
var audioQuality by rememberSaveable { mutableStateOf(AudioQuality.Standard) }
var bufferSize by rememberSaveable { mutableStateOf(8192) }
val analyzer = remember(bufferSize) { SpectrumAnalyzer(bufferSize) }
// spectrumData is a list of pairs of frequency and magnitude,
// you can use it to visualize the frequency spectrum.
var spectrumData by remember { mutableStateOf(FloatArray(0)) }

LaunchedEffect(
    isRecording,
    isPausing,
    audioQuality,
    bufferSize
) {
    if (isRecording && !isPausing) {
        withContext(Dispatchers.Default) {
            val recorder = Kodio.recorder(quality = audioQuality)
            try {
                delay(500) // wait for the recorder to be released
                recorder.start()
                recorder.liveAudioFlow?.toAudioWindows(
                    windowSize = bufferSize,
                    hopSize = bufferSize / 2,
                    channels = audioQuality.format.channels.count,
                    bytesPerSample = audioQuality.format.bytesPerSample
                )?.collect { floatWindow ->
                    spectrumData = analyzer
                        .processAudioWindow(floatWindow)
                }
            } finally {
                recorder.release()
            }
        }
    }
}