@prosopo/audio-assets 0.0.0-stage → 0.2.1

package/dist/cjs/phonemes.cjs

@@ -0,0 +1,217 @@
+//#region src/phonemes.ts
+var STOP_BURST_MS = 18;
+var VOICED_FRICATIVE_VOICE_GAIN = .45;
+var NASAL_VOICE_GAIN = .5;
+var APPROXIMANT_VOICE_GAIN = .85;
+var segmentDefaults = {
+	bandwidths: [
+		80,
+		110,
+		160
+	],
+	voiceGain: 1,
+	noiseGain: 0,
+	noiseCentreHz: 0,
+	noiseBandwidthHz: 0
+};
+var vowel = (id, f1, f2, f3, durationMs) => ({
+	...segmentDefaults,
+	id,
+	excitation: "voiced",
+	formants: [
+		f1,
+		f2,
+		f3
+	],
+	durationMs
+});
+var fricative = (id, centreHz, bandwidthHz, gain, durationMs) => ({
+	...segmentDefaults,
+	id,
+	excitation: "unvoiced",
+	formants: [
+		0,
+		0,
+		0
+	],
+	durationMs,
+	voiceGain: 0,
+	noiseGain: gain,
+	noiseCentreHz: centreHz,
+	noiseBandwidthHz: bandwidthHz
+});
+var voicedFricative = (id, f1, f2, f3, centreHz, bandwidthHz, gain, durationMs) => ({
+	...segmentDefaults,
+	id,
+	excitation: "mixed",
+	formants: [
+		f1,
+		f2,
+		f3
+	],
+	durationMs,
+	voiceGain: VOICED_FRICATIVE_VOICE_GAIN,
+	noiseGain: gain,
+	noiseCentreHz: centreHz,
+	noiseBandwidthHz: bandwidthHz
+});
+var nasal = (id, f1, f2, f3, durationMs) => ({
+	...segmentDefaults,
+	id,
+	excitation: "voiced",
+	formants: [
+		f1,
+		f2,
+		f3
+	],
+	bandwidths: [
+		180,
+		250,
+		320
+	],
+	durationMs,
+	voiceGain: NASAL_VOICE_GAIN
+});
+var closure = (id, durationMs) => ({
+	...segmentDefaults,
+	id,
+	excitation: "silence",
+	formants: [
+		0,
+		0,
+		0
+	],
+	durationMs,
+	voiceGain: 0
+});
+var burst = (id, centreHz, bandwidthHz, gain) => ({
+	...segmentDefaults,
+	id,
+	excitation: "unvoiced",
+	formants: [
+		0,
+		0,
+		0
+	],
+	durationMs: STOP_BURST_MS,
+	voiceGain: 0,
+	noiseGain: gain,
+	noiseCentreHz: centreHz,
+	noiseBandwidthHz: bandwidthHz,
+	abrupt: true
+});
+var IY = vowel("IY", 270, 2290, 3010, 150);
+var IH = vowel("IH", 390, 1990, 2550, 95);
+var EH = vowel("EH", 530, 1840, 2480, 110);
+var AH = vowel("AH", 640, 1190, 2390, 105);
+var AA = vowel("AA", 730, 1090, 2440, 140);
+var AO = vowel("AO", 570, 840, 2410, 150);
+var UW = vowel("UW", 300, 870, 2240, 140);
+var OW_OFF = vowel("OW^", 330, 900, 2300, 90);
+var R = {
+	...vowel("R", 490, 1350, 1690, 85),
+	voiceGain: APPROXIMANT_VOICE_GAIN
+};
+var W = {
+	...vowel("W", 300, 610, 2200, 70),
+	voiceGain: APPROXIMANT_VOICE_GAIN
+};
+var N = nasal("N", 250, 1750, 2600, 90);
+var S = fricative("S", 5800, 3200, .5, 130);
+var F = fricative("F", 4200, 5e3, .24, 120);
+var TH = fricative("TH", 5200, 5600, .2, 110);
+var Z = voicedFricative("Z", 300, 1600, 2500, 5200, 3e3, .3, 110);
+var V = voicedFricative("V", 320, 1100, 2400, 4e3, 4600, .16, 95);
+var T = [closure("T-", 45), burst("T+", 3800, 3400, .55)];
+var K = [closure("K-", 45), burst("K+", 2100, 2200, .5)];
+var AY = [AA, {
+	...IY,
+	durationMs: 110
+}];
+var EY = [EH, {
+	...IY,
+	durationMs: 105
+}];
+/** "zero", not "oh": a lone vowel with no consonant anchor is the first thing lost to noise. */
+var DIGITS = [
+	{
+		answer: "0",
+		phonemes: [
+			Z,
+			IH,
+			R,
+			...[AO, OW_OFF]
+		]
+	},
+	{
+		answer: "1",
+		phonemes: [
+			W,
+			AH,
+			N
+		]
+	},
+	{
+		answer: "2",
+		phonemes: [...T, UW]
+	},
+	{
+		answer: "3",
+		phonemes: [
+			TH,
+			R,
+			IY
+		]
+	},
+	{
+		answer: "4",
+		phonemes: [
+			F,
+			AO,
+			R
+		]
+	},
+	{
+		answer: "5",
+		phonemes: [
+			F,
+			...AY,
+			V
+		]
+	},
+	{
+		answer: "6",
+		phonemes: [
+			S,
+			IH,
+			...K,
+			S
+		]
+	},
+	{
+		answer: "7",
+		phonemes: [
+			S,
+			EH,
+			V,
+			AH,
+			N
+		]
+	},
+	{
+		answer: "8",
+		phonemes: [...EY, ...T]
+	},
+	{
+		answer: "9",
+		phonemes: [
+			N,
+			...AY,
+			N
+		]
+	}
+];
+var ANSWER_ALPHABET = DIGITS.map((d) => d.answer).join("");
+//#endregion
+exports.ANSWER_ALPHABET = ANSWER_ALPHABET;
+exports.DIGITS = DIGITS;

package/dist/cjs/synth.cjs

@@ -0,0 +1,222 @@
+const require_levels = require("./levels.cjs");
+//#region src/synth.ts
+/** Not 8 kHz: /s/, /f/ and /θ/ live at 4-8 kHz and are what separate six, five and three. */
+var SAMPLE_RATE = 16e3;
+var NYQUIST_GUARD_HZ = 100;
+var MIN_FORMANT_BANDWIDTH_HZ = 10;
+var MIN_NOISE_BANDWIDTH_HZ = 100;
+var LEVEL_RAMP_SECONDS = .004;
+var EDGE_FADE_SECONDS = .008;
+var WORD_PEAK = .9;
+var SILENT_RMS = 1e-9;
+var GLOTTAL_OPEN_QUOTIENT = .6;
+var GLOTTAL_SPEED_QUOTIENT = .35;
+var FORMANT_INDICES = [
+	0,
+	1,
+	2
+];
+var randomVoice = (prng) => ({
+	f0Hz: prng.range(85, 210),
+	f0DriftRatio: prng.range(.04, .16),
+	formantScale: prng.range(.86, 1.22),
+	rate: prng.range(.82, 1.24),
+	tiltCornerHz: prng.range(4500, 8e3)
+});
+/** Two-pole formant resonator, normalised to unity DC gain so a cascade stays bounded. */
+var createResonator = (sampleRate) => {
+	let y1 = 0;
+	let y2 = 0;
+	let a = 1;
+	let b = 0;
+	let c = 0;
+	return {
+		setTarget(freqHz, bandwidthHz) {
+			const f = Math.min(Math.max(freqHz, 1), sampleRate / 2 - NYQUIST_GUARD_HZ);
+			const bw = Math.max(bandwidthHz, MIN_FORMANT_BANDWIDTH_HZ);
+			const r = Math.exp(-Math.PI * bw / sampleRate);
+			const theta = 2 * Math.PI * f / sampleRate;
+			b = 2 * r * Math.cos(theta);
+			c = -(r * r);
+			a = 1 - b - c;
+		},
+		process(x) {
+			const y = a * x + b * y1 + c * y2;
+			y2 = y1;
+			y1 = y;
+			return y;
+		}
+	};
+};
+var createOnePoleLowPass = (sampleRate, cutoffHz) => {
+	const dt = 1 / sampleRate;
+	const alpha = dt / (1 / (2 * Math.PI * cutoffHz) + dt);
+	let y = 0;
+	return (x) => {
+		y += alpha * (x - y);
+		return y;
+	};
+};
+/**
+* Lip radiation, a +6 dB/octave differentiator. Without it the glottal
+* pulse's DC and steep rolloff bury F2 and F3, which are what tell vowels
+* apart.
+*/
+var createRadiation = () => {
+	let previous = 0;
+	return (x) => {
+		const y = x - previous;
+		previous = x;
+		return y;
+	};
+};
+/** Rosenberg glottal pulse; a bare impulse train through the formants sounds like a buzzer. */
+var createGlottalSource = (sampleRate) => {
+	const rise = GLOTTAL_OPEN_QUOTIENT * (1 - GLOTTAL_SPEED_QUOTIENT);
+	const fall = GLOTTAL_OPEN_QUOTIENT * GLOTTAL_SPEED_QUOTIENT;
+	let phase = 0;
+	return (f0Hz) => {
+		phase += f0Hz / sampleRate;
+		if (phase >= 1) phase -= 1;
+		if (phase < rise) return .5 * (1 - Math.cos(Math.PI * phase / rise));
+		if (phase < GLOTTAL_OPEN_QUOTIENT) return Math.cos(Math.PI / 2 * ((phase - rise) / fall));
+		return 0;
+	};
+};
+var lerp = (from, to, t) => from + (to - from) * t;
+var isBlendable = (previous, current) => !current.abrupt && !previous.abrupt && previous.excitation !== "silence" && current.excitation !== "silence";
+var buildFrames = (phonemes, voice, sampleRate) => {
+	const frames = [];
+	const segments = [];
+	const scaled = (p) => [
+		p.formants[0] * voice.formantScale,
+		p.formants[1] * voice.formantScale,
+		p.formants[2] * voice.formantScale
+	];
+	let previous;
+	for (const current of phonemes) {
+		const lengthSamples = Math.max(1, Math.round(current.durationMs * voice.rate * sampleRate / 1e3));
+		const from = previous && isBlendable(previous, current) ? previous : current;
+		const blendSamples = from === current ? 0 : Math.floor(lengthSamples / 3);
+		const source = scaled(from);
+		const target = scaled(current);
+		segments.push({
+			start: frames.length,
+			end: frames.length + lengthSamples,
+			targetLevel: current.excitation === "silence" ? 0 : Math.max(current.voiceGain, current.noiseGain)
+		});
+		for (let n = 0; n < lengthSamples; n++) {
+			const t = blendSamples > 0 ? Math.min(1, n / blendSamples) : 1;
+			frames.push({
+				formants: [
+					lerp(source[0], target[0], t),
+					lerp(source[1], target[1], t),
+					lerp(source[2], target[2], t)
+				],
+				bandwidths: [
+					lerp(from.bandwidths[0], current.bandwidths[0], t),
+					lerp(from.bandwidths[1], current.bandwidths[1], t),
+					lerp(from.bandwidths[2], current.bandwidths[2], t)
+				],
+				voiceGain: lerp(from.voiceGain, current.voiceGain, t),
+				noiseGain: lerp(from.noiseGain, current.noiseGain, t),
+				noiseCentreHz: current.noiseCentreHz,
+				noiseBandwidthHz: current.noiseBandwidthHz,
+				silent: current.excitation === "silence"
+			});
+		}
+		previous = current;
+	}
+	return {
+		frames,
+		segments
+	};
+};
+/**
+* Rescale each segment to its target level, ramping the gain across
+* boundaries: a gain step clicks, and a click at every phoneme boundary is a
+* free segmentation cue for an attacker.
+*/
+var applySegmentLevels = (samples, segments, sampleRate) => {
+	const rampLength = Math.max(1, Math.round(sampleRate * LEVEL_RAMP_SECONDS));
+	const gains = segments.map((segment) => {
+		if (segment.targetLevel <= 0) return 0;
+		const level = require_levels.rms(samples.subarray(segment.start, segment.end));
+		return level < SILENT_RMS ? 0 : segment.targetLevel / level;
+	});
+	const track = new Float32Array(samples.length);
+	segments.forEach((segment, i) => {
+		track.fill(gains[i] ?? 0, segment.start, segment.end);
+	});
+	for (let i = 1; i < segments.length; i++) {
+		const segment = segments[i];
+		const gain = gains[i];
+		const previousGain = gains[i - 1];
+		if (!segment || gain === void 0 || previousGain === void 0) continue;
+		const length = Math.min(rampLength, segment.end - segment.start);
+		for (let n = 0; n < length; n++) {
+			const index = segment.start + n;
+			if (index >= track.length) break;
+			const t = .5 * (1 - Math.cos(Math.PI * n / length));
+			track[index] = previousGain + (gain - previousGain) * t;
+		}
+	}
+	for (let n = 0; n < samples.length; n++) samples[n] = (samples[n] ?? 0) * (track[n] ?? 0);
+};
+/** Raised-cosine fade in and out, in place. */
+var applyEdgeFade = (samples, fadeLength) => {
+	const length = Math.min(fadeLength, Math.floor(samples.length / 2));
+	for (let n = 0; n < length; n++) {
+		const gain = .5 * (1 - Math.cos(Math.PI * n / length));
+		const tail = samples.length - 1 - n;
+		samples[n] = (samples[n] ?? 0) * gain;
+		samples[tail] = (samples[tail] ?? 0) * gain;
+	}
+};
+var synthesiseUtterance = (utterance, voice, prng, sampleRate = SAMPLE_RATE) => {
+	const { frames, segments } = buildFrames(utterance.phonemes, voice, sampleRate);
+	const out = new Float32Array(frames.length);
+	const formantFilters = [
+		createResonator(sampleRate),
+		createResonator(sampleRate),
+		createResonator(sampleRate)
+	];
+	const noiseFilter = createResonator(sampleRate);
+	const glottis = createGlottalSource(sampleRate);
+	const radiation = createRadiation();
+	const tilt = createOnePoleLowPass(sampleRate, voice.tiltCornerHz);
+	const f0Start = voice.f0Hz * (1 + voice.f0DriftRatio / 2);
+	const f0End = voice.f0Hz * (1 - voice.f0DriftRatio / 2);
+	const jitterDepth = prng.range(.004, .02);
+	frames.forEach((frame, n) => {
+		if (frame.silent) return;
+		const progress = frames.length > 1 ? n / (frames.length - 1) : 0;
+		const f0 = lerp(f0Start, f0End, progress) * (1 + jitterDepth * (prng.next() - .5));
+		let sample = 0;
+		if (frame.voiceGain > 0) {
+			let voiced = tilt(glottis(f0));
+			for (const k of FORMANT_INDICES) {
+				formantFilters[k].setTarget(frame.formants[k], frame.bandwidths[k]);
+				voiced = formantFilters[k].process(voiced);
+			}
+			sample += voiced * frame.voiceGain;
+		}
+		if (frame.noiseGain > 0) {
+			noiseFilter.setTarget(frame.noiseCentreHz, Math.max(frame.noiseBandwidthHz, MIN_NOISE_BANDWIDTH_HZ));
+			const noise = prng.next() * 2 - 1;
+			sample += noiseFilter.process(noise) * frame.noiseGain;
+		}
+		out[n] = radiation(sample);
+	});
+	applySegmentLevels(out, segments, sampleRate);
+	applyEdgeFade(out, Math.round(sampleRate * EDGE_FADE_SECONDS));
+	require_levels.normalise(out, WORD_PEAK);
+	return {
+		samples: out,
+		sampleRate
+	};
+};
+//#endregion
+exports.SAMPLE_RATE = SAMPLE_RATE;
+exports.randomVoice = randomVoice;
+exports.synthesiseUtterance = synthesiseUtterance;

package/dist/distort.d.ts

@@ -0,0 +1,7 @@
+import type { Prng } from "@prosopo/puzzle-assets";
+import type { AudioBuffer } from "./types.js";
+export declare const mixInto: (target: Float32Array, source: Float32Array, offset: number, gain: number) => void;
+export declare const addNoiseBed: (buffer: AudioBuffer, prng: Prng, snrDb: number) => void;
+export declare const addReverb: (buffer: AudioBuffer, mix: number) => void;
+export declare const finalise: (buffer: AudioBuffer) => void;
+//# sourceMappingURL=distort.d.ts.map

package/dist/distort.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"distort.d.ts","sourceRoot":"","sources":["../src/distort.ts"],"names":[],"mappings":"AAsBA,OAAO,KAAK,EAAE,IAAI,EAAE,MAAM,wBAAwB,CAAC;AAEnD,OAAO,KAAK,EAAE,WAAW,EAAE,MAAM,YAAY,CAAC;AAU9C,eAAO,MAAM,OAAO,WACX,YAAY,UACZ,YAAY,UACZ,MAAM,QACR,MAAM,KACV,IAMF,CAAC;AAMF,eAAO,MAAM,WAAW,WACf,WAAW,QACb,IAAI,SACH,MAAM,KACX,IAsBF,CAAC;AAUF,eAAO,MAAM,SAAS,WAAY,WAAW,OAAO,MAAM,KAAG,IAkC5D,CAAC;AAMF,eAAO,MAAM,QAAQ,WAAY,WAAW,KAAG,IAM9C,CAAC"}

package/dist/distort.js

@@ -0,0 +1,88 @@
+import { normalise, rms } from "./levels.js";
+//#region src/distort.ts
+var REVERB_COMB_DELAYS_MS = [
+	29.7,
+	37.1,
+	41.1,
+	43.7
+];
+var REVERB_COMB_FEEDBACK = .72;
+var REVERB_ALLPASS_DELAYS_MS = [5, 1.7];
+var REVERB_ALLPASS_FEEDBACK = .5;
+var SOFT_CLIP_DRIVE = 1.2;
+var OUTPUT_PEAK = .92;
+/** Adds `source` into `target` from `offset`; anything past the end is dropped. */
+var mixInto = (target, source, offset, gain) => {
+	const start = Math.max(0, Math.floor(offset));
+	const count = Math.min(source.length, target.length - start);
+	for (let n = 0; n < count; n++) target[start + n] = (target[start + n] ?? 0) + (source[n] ?? 0) * gain;
+};
+/**
+* Pink-ish noise at `snrDb` below the signal's RMS. Pink rather than white
+* because white noise spends its energy above the formants and masks little.
+*/
+var addNoiseBed = (buffer, prng, snrDb) => {
+	const signalRms = rms(buffer.samples);
+	if (signalRms === 0) return;
+	const targetRms = signalRms / 10 ** (snrDb / 20);
+	let b0 = 0;
+	let b1 = 0;
+	let b2 = 0;
+	const noise = new Float32Array(buffer.samples.length);
+	for (let n = 0; n < noise.length; n++) {
+		const white = prng.next() * 2 - 1;
+		b0 = .99765 * b0 + white * .099;
+		b1 = .963 * b1 + white * .2965;
+		b2 = .57 * b2 + white * 1.0526;
+		noise[n] = (b0 + b1 + b2 + white * .1848) * .2;
+	}
+	const noiseRms = rms(noise);
+	if (noiseRms === 0) return;
+	mixInto(buffer.samples, noise, 0, targetRms / noiseRms);
+};
+var delaySamples = (delayMs, sampleRate) => Math.max(1, Math.round(delayMs / 1e3 * sampleRate));
+/**
+* Schroeder reverb: parallel combs into series allpasses. Its purpose is to
+* smear onsets, which otherwise make it easy to chop the clip into
+* one-digit segments and classify each in isolation.
+*/
+var addReverb = (buffer, mix) => {
+	if (mix <= 0) return;
+	const { samples, sampleRate } = buffer;
+	const wet = new Float32Array(samples.length);
+	for (const delayMs of REVERB_COMB_DELAYS_MS) {
+		const delay = delaySamples(delayMs, sampleRate);
+		const line = new Float32Array(delay);
+		let index = 0;
+		for (let n = 0; n < samples.length; n++) {
+			const delayed = line[index] ?? 0;
+			wet[n] = (wet[n] ?? 0) + delayed / REVERB_COMB_DELAYS_MS.length;
+			line[index] = (samples[n] ?? 0) + delayed * REVERB_COMB_FEEDBACK;
+			index = (index + 1) % delay;
+		}
+	}
+	for (const delayMs of REVERB_ALLPASS_DELAYS_MS) {
+		const delay = delaySamples(delayMs, sampleRate);
+		const line = new Float32Array(delay);
+		let index = 0;
+		for (let n = 0; n < wet.length; n++) {
+			const input = wet[n] ?? 0;
+			const delayed = line[index] ?? 0;
+			line[index] = input + delayed * REVERB_ALLPASS_FEEDBACK;
+			wet[n] = -input * REVERB_ALLPASS_FEEDBACK + delayed;
+			index = (index + 1) % delay;
+		}
+	}
+	for (let n = 0; n < samples.length; n++) samples[n] = (samples[n] ?? 0) * (1 - mix) + (wet[n] ?? 0) * mix;
+};
+/**
+* tanh soft clip, then normalise: hard clipping at 16-bit conversion is
+* audibly nasty and a distinctive artefact.
+*/
+var finalise = (buffer) => {
+	const { samples } = buffer;
+	for (let n = 0; n < samples.length; n++) samples[n] = Math.tanh((samples[n] ?? 0) * SOFT_CLIP_DRIVE);
+	normalise(samples, OUTPUT_PEAK);
+};
+//#endregion
+export { addNoiseBed, addReverb, finalise, mixInto };

package/dist/distort.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"distort.js","sourceRoot":"","sources":["../src/distort.ts"],"names":[],"mappings":"AAuBA,OAAO,EAAE,SAAS,EAAE,GAAG,EAAE,MAAM,aAAa,CAAC;AAG7C,MAAM,qBAAqB,GAAG,CAAC,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,CAAC,CAAC;AACvD,MAAM,oBAAoB,GAAG,IAAI,CAAC;AAClC,MAAM,wBAAwB,GAAG,CAAC,GAAG,EAAE,GAAG,CAAC,CAAC;AAC5C,MAAM,uBAAuB,GAAG,GAAG,CAAC;AACpC,MAAM,eAAe,GAAG,GAAG,CAAC;AAC5B,MAAM,WAAW,GAAG,IAAI,CAAC;AAGzB,MAAM,CAAC,MAAM,OAAO,GAAG,CACtB,MAAoB,EACpB,MAAoB,EACpB,MAAc,EACd,IAAY,EACL,EAAE;IACT,MAAM,KAAK,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC;IAC9C,MAAM,KAAK,GAAG,IAAI,CAAC,GAAG,CAAC,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,MAAM,GAAG,KAAK,CAAC,CAAC;IAC7D,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,KAAK,EAAE,CAAC,EAAE,EAAE,CAAC;QAChC,MAAM,CAAC,KAAK,GAAG,CAAC,CAAC,GAAG,CAAC,MAAM,CAAC,KAAK,GAAG,CAAC,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,GAAG,IAAI,CAAC;IACxE,CAAC;AACF,CAAC,CAAC;AAMF,MAAM,CAAC,MAAM,WAAW,GAAG,CAC1B,MAAmB,EACnB,IAAU,EACV,KAAa,EACN,EAAE;IACT,MAAM,SAAS,GAAG,GAAG,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC;IACtC,IAAI,SAAS,KAAK,CAAC;QAAE,OAAO;IAE5B,MAAM,SAAS,GAAG,SAAS,GAAG,EAAE,IAAI,CAAC,KAAK,GAAG,EAAE,CAAC,CAAC;IAGjD,IAAI,EAAE,GAAG,CAAC,CAAC;IACX,IAAI,EAAE,GAAG,CAAC,CAAC;IACX,IAAI,EAAE,GAAG,CAAC,CAAC;IACX,MAAM,KAAK,GAAG,IAAI,YAAY,CAAC,MAAM,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC;IACtD,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,KAAK,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QACvC,MAAM,KAAK,GAAG,IAAI,CAAC,IAAI,EAAE,GAAG,CAAC,GAAG,CAAC,CAAC;QAClC,EAAE,GAAG,OAAO,GAAG,EAAE,GAAG,KAAK,GAAG,KAAK,CAAC;QAClC,EAAE,GAAG,KAAK,GAAG,EAAE,GAAG,KAAK,GAAG,MAAM,CAAC;QACjC,EAAE,GAAG,IAAI,GAAG,EAAE,GAAG,KAAK,GAAG,MAAM,CAAC;QAChC,KAAK,CAAC,CAAC,CAAC,GAAG,CAAC,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,KAAK,GAAG,MAAM,CAAC,GAAG,GAAG,CAAC;IAClD,CAAC;IAED,MAAM,QAAQ,GAAG,GAAG,CAAC,KAAK,CAAC,CAAC;IAC5B,IAAI,QAAQ,KAAK,CAAC;QAAE,OAAO;IAC3B,OAAO,CAAC,MAAM,CAAC,OAAO,EAAE,KAAK,EAAE,CAAC,EAAE,SAAS,GAAG,QAAQ,CAAC,CAAC;AACzD,CAAC,CAAC;AAEF,MAAM,YAAY,GAAG,CAAC,OAAe,EAAE,UAAkB,EAAU,EAAE,CACpE,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,KAAK,CAAC,CAAC,OAAO,GAAG,IAAI,CAAC,GAAG,UAAU,CAAC,CAAC,CAAC;AAOxD,MAAM,CAAC,MAAM,SAAS,GAAG,CAAC,MAAmB,EAAE,GAAW,EAAQ,EAAE;IACnE,IAAI,GAAG,IAAI,CAAC;QAAE,OAAO;IAErB,MAAM,EAAE,OAAO,EAAE,UAAU,EAAE,GAAG,MAAM,CAAC;IACvC,MAAM,GAAG,GAAG,IAAI,YAAY,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC;IAE7C,KAAK,MAAM,OAAO,IAAI,qBAAqB,EAAE,CAAC;QAC7C,MAAM,KAAK,GAAG,YAAY,CAAC,OAAO,EAAE,UAAU,CAAC,CAAC;QAChD,MAAM,IAAI,GAAG,IAAI,YAAY,CAAC,KAAK,CAAC,CAAC;QACrC,IAAI,KAAK,GAAG,CAAC,CAAC;QACd,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,OAAO,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;YACzC,MAAM,OAAO,GAAG,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;YACjC,GAAG,CAAC,CAAC,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,GAAG,OAAO,GAAG,qBAAqB,CAAC,MAAM,CAAC;YAChE,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,GAAG,OAAO,GAAG,oBAAoB,CAAC;YACjE,KAAK,GAAG,CAAC,KAAK,GAAG,CAAC,CAAC,GAAG,KAAK,CAAC;QAC7B,CAAC;IACF,CAAC;IAED,KAAK,MAAM,OAAO,IAAI,wBAAwB,EAAE,CAAC;QAChD,MAAM,KAAK,GAAG,YAAY,CAAC,OAAO,EAAE,UAAU,CAAC,CAAC;QAChD,MAAM,IAAI,GAAG,IAAI,YAAY,CAAC,KAAK,CAAC,CAAC;QACrC,IAAI,KAAK,GAAG,CAAC,CAAC;QACd,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,GAAG,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;YACrC,MAAM,KAAK,GAAG,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC;YAC1B,MAAM,OAAO,GAAG,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;YACjC,IAAI,CAAC,KAAK,CAAC,GAAG,KAAK,GAAG,OAAO,GAAG,uBAAuB,CAAC;YACxD,GAAG,CAAC,CAAC,CAAC,GAAG,CAAC,KAAK,GAAG,uBAAuB,GAAG,OAAO,CAAC;YACpD,KAAK,GAAG,CAAC,KAAK,GAAG,CAAC,CAAC,GAAG,KAAK,CAAC;QAC7B,CAAC;IACF,CAAC;IAED,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,OAAO,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QACzC,OAAO,CAAC,CAAC,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,GAAG,GAAG,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,GAAG,GAAG,CAAC;IAClE,CAAC;AACF,CAAC,CAAC;AAMF,MAAM,CAAC,MAAM,QAAQ,GAAG,CAAC,MAAmB,EAAQ,EAAE;IACrD,MAAM,EAAE,OAAO,EAAE,GAAG,MAAM,CAAC;IAC3B,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,OAAO,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QACzC,OAAO,CAAC,CAAC,CAAC,GAAG,IAAI,CAAC,IAAI,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,GAAG,eAAe,CAAC,CAAC;IAC7D,CAAC;IACD,SAAS,CAAC,OAAO,EAAE,WAAW,CAAC,CAAC;AACjC,CAAC,CAAC"}

package/dist/encode.d.ts

@@ -0,0 +1,4 @@
+import type { AudioBuffer } from "./types.js";
+export declare const encodeWav: (buffer: AudioBuffer) => Buffer;
+export declare const toDataUri: (wav: Buffer) => string;
+//# sourceMappingURL=encode.d.ts.map

package/dist/encode.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"encode.d.ts","sourceRoot":"","sources":["../src/encode.ts"],"names":[],"mappings":"AAcA,OAAO,KAAK,EAAE,WAAW,EAAE,MAAM,YAAY,CAAC;AAgB9C,eAAO,MAAM,SAAS,WAAY,WAAW,KAAG,MA+B/C,CAAC;AAEF,eAAO,MAAM,SAAS,QAAS,MAAM,KAAG,MACU,CAAC"}

package/dist/encode.js

@@ -0,0 +1,39 @@
+//#region src/encode.ts
+var RIFF_HEADER_BYTES = 44;
+var RIFF_PREAMBLE_BYTES = 8;
+var FMT_CHUNK_BYTES = 16;
+var BITS_PER_SAMPLE = 16;
+var BYTES_PER_SAMPLE = BITS_PER_SAMPLE / 8;
+var CHANNELS = 1;
+var PCM_FORMAT_TAG = 1;
+var INT16_SCALE = 32767;
+/**
+* Uncompressed WAV because every browser plays 16-bit PCM with no codec
+* dependency or artefacts that could be mistaken for the challenge.
+*/
+var encodeWav = (buffer) => {
+	const { samples, sampleRate } = buffer;
+	const dataBytes = samples.length * BYTES_PER_SAMPLE;
+	const out = Buffer.alloc(RIFF_HEADER_BYTES + dataBytes);
+	out.write("RIFF", 0, "ascii");
+	out.writeUInt32LE(RIFF_HEADER_BYTES - RIFF_PREAMBLE_BYTES + dataBytes, 4);
+	out.write("WAVE", 8, "ascii");
+	out.write("fmt ", 12, "ascii");
+	out.writeUInt32LE(FMT_CHUNK_BYTES, 16);
+	out.writeUInt16LE(PCM_FORMAT_TAG, 20);
+	out.writeUInt16LE(CHANNELS, 22);
+	out.writeUInt32LE(sampleRate, 24);
+	out.writeUInt32LE(sampleRate * CHANNELS * BYTES_PER_SAMPLE, 28);
+	out.writeUInt16LE(CHANNELS * BYTES_PER_SAMPLE, 32);
+	out.writeUInt16LE(BITS_PER_SAMPLE, 34);
+	out.write("data", 36, "ascii");
+	out.writeUInt32LE(dataBytes, 40);
+	for (let n = 0; n < samples.length; n++) {
+		const clamped = Math.max(-1, Math.min(1, samples[n] ?? 0));
+		out.writeInt16LE(Math.round(clamped * INT16_SCALE), RIFF_HEADER_BYTES + n * BYTES_PER_SAMPLE);
+	}
+	return out;
+};
+var toDataUri = (wav) => `data:audio/wav;base64,${wav.toString("base64")}`;
+//#endregion
+export { encodeWav, toDataUri };

package/dist/encode.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"encode.js","sourceRoot":"","sources":["../src/encode.ts"],"names":[],"mappings":"AAgBA,MAAM,iBAAiB,GAAG,EAAE,CAAC;AAC7B,MAAM,mBAAmB,GAAG,CAAC,CAAC;AAC9B,MAAM,eAAe,GAAG,EAAE,CAAC;AAC3B,MAAM,eAAe,GAAG,EAAE,CAAC;AAC3B,MAAM,gBAAgB,GAAG,eAAe,GAAG,CAAC,CAAC;AAC7C,MAAM,QAAQ,GAAG,CAAC,CAAC;AACnB,MAAM,cAAc,GAAG,CAAC,CAAC;AAEzB,MAAM,WAAW,GAAG,MAAM,CAAC;AAM3B,MAAM,CAAC,MAAM,SAAS,GAAG,CAAC,MAAmB,EAAU,EAAE;IACxD,MAAM,EAAE,OAAO,EAAE,UAAU,EAAE,GAAG,MAAM,CAAC;IACvC,MAAM,SAAS,GAAG,OAAO,CAAC,MAAM,GAAG,gBAAgB,CAAC;IACpD,MAAM,GAAG,GAAG,MAAM,CAAC,KAAK,CAAC,iBAAiB,GAAG,SAAS,CAAC,CAAC;IAExD,GAAG,CAAC,KAAK,CAAC,MAAM,EAAE,CAAC,EAAE,OAAO,CAAC,CAAC;IAC9B,GAAG,CAAC,aAAa,CAAC,iBAAiB,GAAG,mBAAmB,GAAG,SAAS,EAAE,CAAC,CAAC,CAAC;IAC1E,GAAG,CAAC,KAAK,CAAC,MAAM,EAAE,CAAC,EAAE,OAAO,CAAC,CAAC;IAE9B,GAAG,CAAC,KAAK,CAAC,MAAM,EAAE,EAAE,EAAE,OAAO,CAAC,CAAC;IAC/B,GAAG,CAAC,aAAa,CAAC,eAAe,EAAE,EAAE,CAAC,CAAC;IACvC,GAAG,CAAC,aAAa,CAAC,cAAc,EAAE,EAAE,CAAC,CAAC;IACtC,GAAG,CAAC,aAAa,CAAC,QAAQ,EAAE,EAAE,CAAC,CAAC;IAChC,GAAG,CAAC,aAAa,CAAC,UAAU,EAAE,EAAE,CAAC,CAAC;IAClC,GAAG,CAAC,aAAa,CAAC,UAAU,GAAG,QAAQ,GAAG,gBAAgB,EAAE,EAAE,CAAC,CAAC;IAChE,GAAG,CAAC,aAAa,CAAC,QAAQ,GAAG,gBAAgB,EAAE,EAAE,CAAC,CAAC;IACnD,GAAG,CAAC,aAAa,CAAC,eAAe,EAAE,EAAE,CAAC,CAAC;IAEvC,GAAG,CAAC,KAAK,CAAC,MAAM,EAAE,EAAE,EAAE,OAAO,CAAC,CAAC;IAC/B,GAAG,CAAC,aAAa,CAAC,SAAS,EAAE,EAAE,CAAC,CAAC;IAEjC,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,OAAO,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QAEzC,MAAM,OAAO,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,OAAO,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;QAC3D,GAAG,CAAC,YAAY,CACf,IAAI,CAAC,KAAK,CAAC,OAAO,GAAG,WAAW,CAAC,EACjC,iBAAiB,GAAG,CAAC,GAAG,gBAAgB,CACxC,CAAC;IACH,CAAC;IAED,OAAO,GAAG,CAAC;AACZ,CAAC,CAAC;AAEF,MAAM,CAAC,MAAM,SAAS,GAAG,CAAC,GAAW,EAAU,EAAE,CAChD,yBAAyB,GAAG,CAAC,QAAQ,CAAC,QAAQ,CAAC,EAAE,CAAC"}

package/dist/index.d.ts

@@ -0,0 +1,13 @@
+import type { AudioRenderSettings, RenderedAudioChallenge } from "./types.js";
+export { createPrng, createSeed } from "@prosopo/puzzle-assets";
+export type { Prng } from "@prosopo/puzzle-assets";
+export { ANSWER_ALPHABET, DIGITS } from "./phonemes.js";
+export { SAMPLE_RATE, randomVoice, synthesiseUtterance } from "./synth.js";
+export type { Voice } from "./synth.js";
+export { encodeWav, toDataUri } from "./encode.js";
+export { addNoiseBed } from "./distort.js";
+export { rms } from "./levels.js";
+export type { AudioBuffer, AudioRenderSettings, Excitation, Phoneme, RenderedAudioChallenge, Utterance, } from "./types.js";
+export declare const DEFAULT_RENDER_SETTINGS: AudioRenderSettings;
+export declare const renderAudioChallenge: (settings?: AudioRenderSettings, sampleRate?: number) => RenderedAudioChallenge;
+//# sourceMappingURL=index.d.ts.map

package/dist/index.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../src/index.ts"],"names":[],"mappings":"AAwBA,OAAO,KAAK,EAEX,mBAAmB,EACnB,sBAAsB,EAEtB,MAAM,YAAY,CAAC;AAEpB,OAAO,EAAE,UAAU,EAAE,UAAU,EAAE,MAAM,wBAAwB,CAAC;AAChE,YAAY,EAAE,IAAI,EAAE,MAAM,wBAAwB,CAAC;AACnD,OAAO,EAAE,eAAe,EAAE,MAAM,EAAE,MAAM,eAAe,CAAC;AACxD,OAAO,EAAE,WAAW,EAAE,WAAW,EAAE,mBAAmB,EAAE,MAAM,YAAY,CAAC;AAC3E,YAAY,EAAE,KAAK,EAAE,MAAM,YAAY,CAAC;AACxC,OAAO,EAAE,SAAS,EAAE,SAAS,EAAE,MAAM,aAAa,CAAC;AACnD,OAAO,EAAE,WAAW,EAAE,MAAM,cAAc,CAAC;AAC3C,OAAO,EAAE,GAAG,EAAE,MAAM,aAAa,CAAC;AAClC,YAAY,EACX,WAAW,EACX,mBAAmB,EACnB,UAAU,EACV,OAAO,EACP,sBAAsB,EACtB,SAAS,GACT,MAAM,YAAY,CAAC;AAGpB,eAAO,MAAM,uBAAuB,EAAE,mBAOrC,CAAC;AAuDF,eAAO,MAAM,oBAAoB,cACtB,mBAAmB,eACjB,MAAM,KAChB,sBA6DF,CAAC"}