frontend/js/audio-processor.js

/**
 * AudioWorklet Processor for exact chunk size control
 * Accumulates audio samples and sends exactly 1600 samples (100ms at 16kHz) per chunk
 *
 * IMPORTANT: This processor resamples from browser's native sample rate (usually 48kHz)
 * down to 16kHz required by the backend.
 */

class AudioChunkProcessor extends AudioWorkletProcessor {
    constructor() {
        super();
        this.buffer = [];
        // Target: 100ms at 16kHz = 1600 samples (matches VAD chunk_duration_ms)
        // Server accumulates 6x 100ms chunks into 600ms chunks for ASR
        this.targetSamples = 1600;
        this.targetSampleRate = 16000;

        // Resampling state
        this.resampleBuffer = [];
        this.inputSampleRate = sampleRate; // AudioContext's native sample rate (e.g., 48000)
        this.resampleRatio = this.inputSampleRate / this.targetSampleRate; // e.g., 48000/16000 = 3
    }

    process(inputs, outputs, parameters) {
        const input = inputs[0];
        if (!input || !input[0]) return true;

        const inputData = input[0]; // Float32Array from Web Audio API (at native sample rate, e.g., 48kHz)

        // Resample from native rate (48kHz) to target rate (16kHz)
        // Using simple linear interpolation decimation
        for (let i = 0; i < inputData.length; i++) {
            this.resampleBuffer.push(inputData[i]);
        }

        // Decimate: take every Nth sample (e.g., every 3rd sample for 48kHz->16kHz)
        while (this.resampleBuffer.length >= this.resampleRatio) {
            // Simple decimation: take first sample, discard rest
            // For better quality, could use averaging or proper low-pass filter
            const sample = this.resampleBuffer[0];
            this.buffer.push(sample);

            // Remove processed samples
            this.resampleBuffer.splice(0, Math.floor(this.resampleRatio));
        }

        // Send chunks when we have enough samples
        while (this.buffer.length >= this.targetSamples) {
            const chunk = this.buffer.splice(0, this.targetSamples);

            // Convert Float32Array to Int16Array (PCM 16-bit)
            const int16Data = new Int16Array(chunk.length);
            for (let i = 0; i < chunk.length; i++) {
                const s = Math.max(-1, Math.min(1, chunk[i]));
                int16Data[i] = s < 0 ? s * 0x8000 : s * 0x7FFF;
            }

            // Send to main thread (transfer ownership for zero-copy)
            this.port.postMessage({
                type: 'audio-chunk',
                data: int16Data.buffer,
                samples: this.targetSamples,
                timestamp: currentTime
            }, [int16Data.buffer]);
        }

        return true; // Keep processor alive
    }
}

registerProcessor('audio-chunk-processor', AudioChunkProcessor);