Wav Recorder is a lightweight Kotlin Multiplatform library that captures audio from the system microphone and returns it as a WAV-formatted kotlinx.io.Buffer. It focuses on predictable WAV output for speech pipelines and AI-driven applications without bundling higher-level speech tooling.

• Unified API – obtain a Recorder instance via platformRecorder, start/stop, and receive a WAV buffer you can persist or stream.
• Cross-platform implementations – Android (API 24+), Windows, and macOS ship with working backends; other native targets currently fall back to isAvailable = false.
• Built-in WAV conversion – raw PCM data is converted to canonical 16-bit, mono WAV so you can drop it straight into transcription workflows.
• Extensible error hooks – implement WavRecorderEventHandler to observe initialisation and recording failures (useful for UI messaging or analytics).

kotlin {
    sourceSets {
        commonMain.dependencies {
            implementation("de.findusl:wav-recorder:0.1.1")
        }
    }
}

Ensure mavenCentral() is present in your repository list.

import de.findusl.wavrecorder.platformRecorder
import kotlinx.io.readByteArray

val recorder = platformRecorder
require(recorder.isAvailable) { "Recording is not supported on this device." }

recorder.startRecording()
// ... wait or collect input ...
val wavBuffer = recorder.stopRecording().getOrThrow()
val wavBytes: ByteArray = wavBuffer.readByteArray()
recorder.close()

• Android callers must request android.permission.RECORD_AUDIO before invoking startRecording().
• Always call close() when you no longer need the recorder to release microphone resources.

Implement WavRecorderEventHandler (assign to WavRecorderEventHandler.INSTANCE) to surface issues such as Windows initialisation problems or runtime recording failures:

WavRecorderEventHandler.INSTANCE = object : WavRecorderEventHandler {
    override fun failedToInitializeOnWindows(e: Exception) {
        // Show a toast, log to analytics, etc.
    }

    override fun errorWhileRecording(e: Exception) { /* ... */ }
}

• Android – Uses AudioRecord, dynamically picks a supported sample rate, outputs mono, 16-bit WAV.
• Windows – Relies on javax.sound.sampled.TargetDataLine with a 44.1 kHz mono configuration.
• macOS – Scans available audio devices and selects the first microphone-capable line with a specified sample rate.
• Linux / other native targets – Placeholder implementation; isAvailable returns false until native backends are added.

Planned improvements include native backends (Linux/iOS) and higher-level streaming helpers. Issues and pull requests are welcome—check the existing workflows in .github/workflows if you plan to extend CI or publishing.

This project is licensed under the MIT License.