// Estimate duration based on text and settings
const effectiveRate = (velocityRate + genderRateAdd) * moodRateMulti;
const estimatedDuration = Math.max(1, textLength / (14 * effectiveRate));

// Set up offline audio context for rendering
const sampleRate = 22050;
const numSamples = Math.ceil(estimatedDuration * sampleRate);
const audioCtx = new OfflineAudioContext(1, numSamples, sampleRate);

// Calculate effective pitch from voice type and mood
const effectivePitch = voiceTypePitch * moodPitch;
const baseFreq = 120 * effectivePitch;

// Create sawtooth oscillator
const osc = audioCtx.createOscillator();
osc.type = 'sawtooth';
osc.frequency.setValueAtTime(baseFreq, 0);

// Simulate natural voice variation
for (let i = 0; i < duration; i += 0.3) {
  const variation = (Math.random() - 0.5) * baseFreq * 0.12;
  osc.frequency.linearRampToValueAtTime(baseFreq + variation, i + 0.15);
  osc.frequency.linearRampToValueAtTime(baseFreq, i + 0.3);
}

// Create formant filters for vowel-like sounds
const f1 = audioCtx.createBiquadFilter();
f1.type = 'bandpass';
f1.frequency.value = 800 * effectivePitch;  // First formant
f1.Q.value = 3;

const f2 = audioCtx.createBiquadFilter();
f2.type = 'bandpass';
f2.frequency.value = 2200 * effectivePitch; // Second formant
f2.Q.value = 4;

// Syllable-based volume modulation
const words = text.split(' ');
const wordDuration = estimatedDuration / words.length;

words.forEach((word, index) => {
  const syllables = Math.max(1, Math.ceil(word.length / 3));
  for (let s = 0; s < syllables; s++) {
    const time = index * wordDuration + s * (wordDuration / syllables);
    gainNode.gain.linearRampToValueAtTime(0.18 * moodVolume, time + 0.02);
    gainNode.gain.linearRampToValueAtTime(0.05, time + wordDuration/syllables - 0.02);
  }
});

// Create noise buffer for breathiness
const noiseBuffer = audioCtx.createBuffer(1, numSamples, sampleRate);
const noiseData = noiseBuffer.getChannelData(0);
for (let i = 0; i < numSamples; i++) {
  noiseData[i] = (Math.random() * 2 - 1) * 0.04;
}

// High-pass filter for breath-like quality
const noiseFilter = audioCtx.createBiquadFilter();
noiseFilter.type = 'bandpass';
noiseFilter.frequency.value = 4000;
noiseFilter.Q.value = 2;

// Connect all nodes and render
osc.connect(f1);
osc.connect(f2);
f1.connect(gainNode);
f2.connect(gainNode);
noiseSource.connect(noiseFilter);
noiseFilter.connect(gainNode);
gainNode.connect(audioCtx.destination);

// Start sources and render
osc.start(0);
osc.stop(estimatedDuration);
noiseSource.start(0);
noiseSource.stop(estimatedDuration);

const renderedBuffer = await audioCtx.startRendering();

// Extract samples and encode to WAV
const channelData = renderedBuffer.getChannelData(0);
const wavBuffer = encodeWAV(channelData, sampleRate);

// Create downloadable blob
const blob = new Blob([wavBuffer], { type: 'audio/wav' });
const url = URL.createObjectURL(blob);

// Trigger download
const a = document.createElement('a');
a.href = url;
a.download = `vozcraft-${item.id}.${format}`;
a.click();