feat: scaffold realtime Korean voice assistant bot

src/audio/pcm.ts

@@ -0,0 +1,60 @@
+export class Stereo48kToMono16kDownsampler {
+  private readonly pendingMono48k: number[] = [];
+
+  pushStereo48kChunk(chunk: Buffer): Int16Array {
+    if (chunk.length < 4) {
+      return new Int16Array();
+    }
+
+    for (let offset = 0; offset + 3 < chunk.length; offset += 4) {
+      const left = chunk.readInt16LE(offset);
+      const right = chunk.readInt16LE(offset + 2);
+      this.pendingMono48k.push(Math.round((left + right) / 2));
+    }
+
+    const outputLength = Math.floor(this.pendingMono48k.length / 3);
+    if (outputLength === 0) {
+      return new Int16Array();
+    }
+
+    const output = new Int16Array(outputLength);
+    let readIndex = 0;
+    for (let index = 0; index < outputLength; index += 1) {
+      const a = this.pendingMono48k[readIndex];
+      const b = this.pendingMono48k[readIndex + 1];
+      const c = this.pendingMono48k[readIndex + 2];
+      output[index] = Math.round((a + b + c) / 3);
+      readIndex += 3;
+    }
+
+    this.pendingMono48k.splice(0, readIndex);
+    return output;
+  }
+}
+
+export function int16ArrayToFloat32(input: Int16Array): Float32Array {
+  const output = new Float32Array(input.length);
+  for (let index = 0; index < input.length; index += 1) {
+    output[index] = input[index] / 32768;
+  }
+  return output;
+}
+
+export function float32ToPcm16Buffer(input: Float32Array): Buffer {
+  const buffer = Buffer.allocUnsafe(input.length * 2);
+  for (let index = 0; index < input.length; index += 1) {
+    const value = Math.max(-1, Math.min(1, input[index]));
+    const scaled = value < 0 ? value * 32768 : value * 32767;
+    buffer.writeInt16LE(Math.round(scaled), index * 2);
+  }
+  return buffer;
+}
+
+export function takeFrame(source: number[], frameSize: number): Int16Array | null {
+  if (source.length < frameSize) {
+    return null;
+  }
+
+  const values = source.splice(0, frameSize);
+  return Int16Array.from(values);
+}