@speech-sdk/core 0.5.2 → 0.6.1

package/dist/conversation/pcm-concat.js

@@ -0,0 +1,246 @@
+import { parseMediaTypeParam, wrapPcm16Mono } from "../audio-utils.js";
+/**
+ * View 16-bit little-endian PCM bytes as an Int16Array. Reuses the existing
+ * buffer when `byteOffset` is 2-aligned; otherwise copies into a fresh,
+ * aligned buffer (Int16Array's buffer view requires 2-byte alignment).
+ */
+function pcmBytesToInt16(bytes) {
+    if (bytes.byteOffset % 2 === 0 && bytes.byteLength % 2 === 0) {
+        return new Int16Array(bytes.buffer, bytes.byteOffset, bytes.byteLength / 2);
+    }
+    const copy = new Uint8Array(bytes.byteLength);
+    copy.set(bytes);
+    return new Int16Array(copy.buffer);
+}
+function downmixToMono(interleaved, channels) {
+    if (channels === 1) {
+        return interleaved;
+    }
+    const frames = Math.floor(interleaved.length / channels);
+    const out = new Int16Array(frames);
+    for (let f = 0; f < frames; f++) {
+        let sum = 0;
+        for (let c = 0; c < channels; c++) {
+            sum += interleaved[f * channels + c];
+        }
+        out[f] = Math.round(sum / channels);
+    }
+    return out;
+}
+/**
+ * View 32-bit little-endian float PCM bytes as a Float32Array. Reuses the
+ * existing buffer when 4-aligned; otherwise copies into a fresh, aligned
+ * buffer (Float32Array's view requires 4-byte alignment).
+ */
+function pcmBytesToFloat32(bytes) {
+    if (bytes.byteOffset % 4 === 0 && bytes.byteLength % 4 === 0) {
+        return new Float32Array(bytes.buffer, bytes.byteOffset, bytes.byteLength / 4);
+    }
+    const copy = new Uint8Array(bytes.byteLength);
+    copy.set(bytes);
+    return new Float32Array(copy.buffer);
+}
+const INT16_MAX = 32_767;
+const INT16_MIN = -32_768;
+/** Convert normalized [-1,1] float32 samples to int16 with clamping. */
+function float32ToInt16(f32) {
+    const out = new Int16Array(f32.length);
+    for (let i = 0; i < f32.length; i++) {
+        const s = f32[i];
+        if (s >= 1) {
+            out[i] = INT16_MAX;
+        }
+        else if (s <= -1) {
+            out[i] = INT16_MIN;
+        }
+        else {
+            out[i] = Math.round(s * INT16_MAX);
+        }
+    }
+    return out;
+}
+/** Match `encoding=<value>` (string) in a mediaType param list. */
+const ENCODING_PARAM_RE = /(?:^|;)\s*encoding=([a-z0-9_-]+)(?=$|;|\s)/i;
+/** Decode a provider response to mono 16-bit PCM + its native sample rate. */
+export function decodeToPcm16(data, mediaType) {
+    const lower = mediaType.toLowerCase();
+    if (lower.startsWith("audio/pcm") || lower.startsWith("audio/x-pcm")) {
+        // NOTE: `audio/l16` (RFC 2586) is intentionally NOT handled here. The
+        // standard mandates network byte order (big-endian) but no provider in
+        // this SDK currently emits it. If support is added later, byte-swap on
+        // little-endian hosts before constructing the Int16Array.
+        const sampleRate = parseMediaTypeParam(mediaType, "rate") ?? 24_000;
+        const channels = parseMediaTypeParam(mediaType, "channels") ?? 1;
+        const encoding = lower.match(ENCODING_PARAM_RE)?.[1];
+        if (encoding === "float32") {
+            const interleaved = float32ToInt16(pcmBytesToFloat32(data));
+            return {
+                pcm: downmixToMono(interleaved, channels),
+                sampleRate,
+                channels: 1,
+            };
+        }
+        const interleaved = pcmBytesToInt16(data);
+        return {
+            pcm: downmixToMono(interleaved, channels),
+            sampleRate,
+            channels: 1,
+        };
+    }
+    if (lower.startsWith("audio/wav") || lower.startsWith("audio/x-wav")) {
+        return decodeWav(data);
+    }
+    throw new Error(`conversation.pcm-concat: unsupported stitch mediaType "${mediaType}". ` +
+        'getStitchOptions must return "audio/wav" or "audio/pcm;rate=...[;encoding=float32]" so the stitch layer can concatenate without a compressed-audio decoder.');
+}
+function decodeWav(bytes) {
+    const view = new DataView(bytes.buffer, bytes.byteOffset, bytes.byteLength);
+    if (view.getUint32(0) !== 0x52_49_46_46 ||
+        view.getUint32(8) !== 0x57_41_56_45) {
+        throw new Error("conversation.pcm-concat: not a RIFF/WAVE file");
+    }
+    // Scan chunks for "fmt " and "data".
+    let offset = 12;
+    let sampleRate = 0;
+    let channels = 0;
+    let bitsPerSample = 0;
+    let audioFormat = 0;
+    let dataStart = -1;
+    let dataLen = 0;
+    while (offset + 8 <= bytes.byteLength) {
+        const chunkId = view.getUint32(offset);
+        const chunkSize = view.getUint32(offset + 4, true);
+        if (chunkId === 0x66_6d_74_20) {
+            audioFormat = view.getUint16(offset + 8, true);
+            channels = view.getUint16(offset + 10, true);
+            sampleRate = view.getUint32(offset + 12, true);
+            bitsPerSample = view.getUint16(offset + 22, true);
+        }
+        else if (chunkId === 0x64_61_74_61) {
+            dataStart = offset + 8;
+            dataLen = chunkSize;
+            break;
+        }
+        offset += 8 + chunkSize + (chunkSize % 2);
+    }
+    if (dataStart < 0 ||
+        sampleRate === 0 ||
+        bitsPerSample !== 16 ||
+        audioFormat !== 1) {
+        throw new Error(`conversation.pcm-concat: only 16-bit PCM WAV is supported (got audioFormat=${audioFormat}, bps=${bitsPerSample})`);
+    }
+    const payload = bytes.subarray(dataStart, dataStart + dataLen);
+    const interleaved = pcmBytesToInt16(payload);
+    return {
+        pcm: downmixToMono(interleaved, channels || 1),
+        sampleRate,
+        channels: 1,
+    };
+}
+/** Simple linear interpolation resampler for mono Int16 PCM. */
+function resamplePcm16LinearMono(input, fromRate, toRate) {
+    if (fromRate === toRate) {
+        return input;
+    }
+    const ratio = fromRate / toRate;
+    const outLen = Math.round(input.length / ratio);
+    const out = new Int16Array(outLen);
+    for (let i = 0; i < outLen; i++) {
+        const srcPos = i * ratio;
+        const i0 = Math.floor(srcPos);
+        const i1 = Math.min(i0 + 1, input.length - 1);
+        const frac = srcPos - i0;
+        out[i] = Math.round(input[i0] * (1 - frac) + input[i1] * frac);
+    }
+    return out;
+}
+function silencePcm16(ms, sampleRate) {
+    const samples = Math.round((ms / 1000) * sampleRate);
+    return new Int16Array(samples);
+}
+/** Root-mean-square amplitude of a PCM segment. */
+function rmsPcm16(pcm) {
+    if (pcm.length === 0) {
+        return 0;
+    }
+    let sumSq = 0;
+    for (const s of pcm) {
+        sumSq += s * s;
+    }
+    return Math.sqrt(sumSq / pcm.length);
+}
+function clampInt16(value) {
+    if (value > INT16_MAX) {
+        return INT16_MAX;
+    }
+    if (value < INT16_MIN) {
+        return INT16_MIN;
+    }
+    return value;
+}
+/** Multiply each sample by `gain`, clamping to int16 range. */
+function scaleClamp(pcm, gain) {
+    const out = new Int16Array(pcm.length);
+    for (let i = 0; i < pcm.length; i++) {
+        out[i] = clampInt16(Math.round(pcm[i] * gain));
+    }
+    return out;
+}
+/**
+ * Default RMS target: −20 dBFS — broadcast/podcast voice loudness convention
+ * with ~20 dB peak headroom. Comfortable to listen to and leaves room for
+ * typical TTS peaks not to clip.
+ */
+export const DEFAULT_VOLUME_DBFS = -20;
+/** Convert a dBFS level (≤ 0) to the equivalent int16 RMS amplitude. */
+export function dbfsToInt16Rms(dbfs) {
+    return Math.round(INT16_MAX * 10 ** (dbfs / 20));
+}
+const DEFAULT_TARGET_RMS_INT16 = dbfsToInt16Rms(DEFAULT_VOLUME_DBFS);
+/**
+ * RMS-normalize each segment to an absolute target amplitude. Each segment
+ * is processed independently — no cross-segment dependency — so:
+ *   - The output level is the same across runs regardless of input mix.
+ *   - Two `generateConversation` calls produce comparable loudness even
+ *     with completely different content.
+ *
+ * Silent segments pass through unchanged. Output is clamped to int16
+ * range, so a quiet segment with rare peaks may clip slightly when
+ * boosted; the default target leaves ~20 dB headroom to make this rare
+ * for typical TTS content.
+ */
+export function normalizeRms(segments, targetRmsAmplitude = DEFAULT_TARGET_RMS_INT16) {
+    return segments.map((s) => {
+        const segRms = rmsPcm16(s.pcm);
+        if (segRms === 0) {
+            return { ...s };
+        }
+        return { ...s, pcm: scaleClamp(s.pcm, targetRmsAmplitude / segRms) };
+    });
+}
+/**
+ * Resample each segment to `targetSampleRate` mono, interleave with `gapMs`
+ * silence, and mux the result as a WAV file via mediabunny.
+ */
+export async function concatPcmToWav(segments, options) {
+    const { gapMs, targetSampleRate } = options;
+    const resampled = [];
+    const gap = silencePcm16(gapMs, targetSampleRate);
+    for (let i = 0; i < segments.length; i++) {
+        const s = segments[i];
+        resampled.push(resamplePcm16LinearMono(s.pcm, s.sampleRate, targetSampleRate));
+        if (i < segments.length - 1 && gap.length > 0) {
+            resampled.push(gap);
+        }
+    }
+    const totalSamples = resampled.reduce((n, a) => n + a.length, 0);
+    const merged = new Int16Array(totalSamples);
+    let off = 0;
+    for (const a of resampled) {
+        merged.set(a, off);
+        off += a.length;
+    }
+    const mergedBytes = new Uint8Array(merged.buffer, merged.byteOffset, merged.byteLength);
+    return await wrapPcm16Mono(mergedBytes, targetSampleRate);
+}
+//# sourceMappingURL=pcm-concat.js.map

package/dist/conversation/pcm-concat.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"pcm-concat.js","sourceRoot":"","sources":["../../src/conversation/pcm-concat.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,mBAAmB,EAAE,aAAa,EAAE,MAAM,mBAAmB,CAAC;AAQvE;;;;GAIG;AACH,SAAS,eAAe,CAAC,KAAiB;IACxC,IAAI,KAAK,CAAC,UAAU,GAAG,CAAC,KAAK,CAAC,IAAI,KAAK,CAAC,UAAU,GAAG,CAAC,KAAK,CAAC,EAAE,CAAC;QAC7D,OAAO,IAAI,UAAU,CAAC,KAAK,CAAC,MAAM,EAAE,KAAK,CAAC,UAAU,EAAE,KAAK,CAAC,UAAU,GAAG,CAAC,CAAC,CAAC;IAC9E,CAAC;IACD,MAAM,IAAI,GAAG,IAAI,UAAU,CAAC,KAAK,CAAC,UAAU,CAAC,CAAC;IAC9C,IAAI,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC;IAChB,OAAO,IAAI,UAAU,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;AACrC,CAAC;AAED,SAAS,aAAa,CAAC,WAAuB,EAAE,QAAgB;IAC9D,IAAI,QAAQ,KAAK,CAAC,EAAE,CAAC;QACnB,OAAO,WAAW,CAAC;IACrB,CAAC;IACD,MAAM,MAAM,GAAG,IAAI,CAAC,KAAK,CAAC,WAAW,CAAC,MAAM,GAAG,QAAQ,CAAC,CAAC;IACzD,MAAM,GAAG,GAAG,IAAI,UAAU,CAAC,MAAM,CAAC,CAAC;IACnC,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QAChC,IAAI,GAAG,GAAG,CAAC,CAAC;QACZ,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,QAAQ,EAAE,CAAC,EAAE,EAAE,CAAC;YAClC,GAAG,IAAI,WAAW,CAAC,CAAC,GAAG,QAAQ,GAAG,CAAC,CAAC,CAAC;QACvC,CAAC;QACD,GAAG,CAAC,CAAC,CAAC,GAAG,IAAI,CAAC,KAAK,CAAC,GAAG,GAAG,QAAQ,CAAC,CAAC;IACtC,CAAC;IACD,OAAO,GAAG,CAAC;AACb,CAAC;AAED;;;;GAIG;AACH,SAAS,iBAAiB,CAAC,KAAiB;IAC1C,IAAI,KAAK,CAAC,UAAU,GAAG,CAAC,KAAK,CAAC,IAAI,KAAK,CAAC,UAAU,GAAG,CAAC,KAAK,CAAC,EAAE,CAAC;QAC7D,OAAO,IAAI,YAAY,CACrB,KAAK,CAAC,MAAM,EACZ,KAAK,CAAC,UAAU,EAChB,KAAK,CAAC,UAAU,GAAG,CAAC,CACrB,CAAC;IACJ,CAAC;IACD,MAAM,IAAI,GAAG,IAAI,UAAU,CAAC,KAAK,CAAC,UAAU,CAAC,CAAC;IAC9C,IAAI,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC;IAChB,OAAO,IAAI,YAAY,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;AACvC,CAAC;AAED,MAAM,SAAS,GAAG,MAAM,CAAC;AACzB,MAAM,SAAS,GAAG,CAAC,MAAM,CAAC;AAE1B,wEAAwE;AACxE,SAAS,cAAc,CAAC,GAAiB;IACvC,MAAM,GAAG,GAAG,IAAI,UAAU,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC;IACvC,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,GAAG,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QACpC,MAAM,CAAC,GAAG,GAAG,CAAC,CAAC,CAAC,CAAC;QACjB,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;YACX,GAAG,CAAC,CAAC,CAAC,GAAG,SAAS,CAAC;QACrB,CAAC;aAAM,IAAI,CAAC,IAAI,CAAC,CAAC,EAAE,CAAC;YACnB,GAAG,CAAC,CAAC,CAAC,GAAG,SAAS,CAAC;QACrB,CAAC;aAAM,CAAC;YACN,GAAG,CAAC,CAAC,CAAC,GAAG,IAAI,CAAC,KAAK,CAAC,CAAC,GAAG,SAAS,CAAC,CAAC;QACrC,CAAC;IACH,CAAC;IACD,OAAO,GAAG,CAAC;AACb,CAAC;AAED,mEAAmE;AACnE,MAAM,iBAAiB,GAAG,6CAA6C,CAAC;AAExE,8EAA8E;AAC9E,MAAM,UAAU,aAAa,CAC3B,IAAgB,EAChB,SAAiB;IAEjB,MAAM,KAAK,GAAG,SAAS,CAAC,WAAW,EAAE,CAAC;IAEtC,IAAI,KAAK,CAAC,UAAU,CAAC,WAAW,CAAC,IAAI,KAAK,CAAC,UAAU,CAAC,aAAa,CAAC,EAAE,CAAC;QACrE,sEAAsE;QACtE,uEAAuE;QACvE,uEAAuE;QACvE,0DAA0D;QAC1D,MAAM,UAAU,GAAG,mBAAmB,CAAC,SAAS,EAAE,MAAM,CAAC,IAAI,MAAM,CAAC;QACpE,MAAM,QAAQ,GAAG,mBAAmB,CAAC,SAAS,EAAE,UAAU,CAAC,IAAI,CAAC,CAAC;QACjE,MAAM,QAAQ,GAAG,KAAK,CAAC,KAAK,CAAC,iBAAiB,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC;QAErD,IAAI,QAAQ,KAAK,SAAS,EAAE,CAAC;YAC3B,MAAM,WAAW,GAAG,cAAc,CAAC,iBAAiB,CAAC,IAAI,CAAC,CAAC,CAAC;YAC5D,OAAO;gBACL,GAAG,EAAE,aAAa,CAAC,WAAW,EAAE,QAAQ,CAAC;gBACzC,UAAU;gBACV,QAAQ,EAAE,CAAC;aACZ,CAAC;QACJ,CAAC;QAED,MAAM,WAAW,GAAG,eAAe,CAAC,IAAI,CAAC,CAAC;QAC1C,OAAO;YACL,GAAG,EAAE,aAAa,CAAC,WAAW,EAAE,QAAQ,CAAC;YACzC,UAAU;YACV,QAAQ,EAAE,CAAC;SACZ,CAAC;IACJ,CAAC;IAED,IAAI,KAAK,CAAC,UAAU,CAAC,WAAW,CAAC,IAAI,KAAK,CAAC,UAAU,CAAC,aAAa,CAAC,EAAE,CAAC;QACrE,OAAO,SAAS,CAAC,IAAI,CAAC,CAAC;IACzB,CAAC;IAED,MAAM,IAAI,KAAK,CACb,0DAA0D,SAAS,KAAK;QACtE,6JAA6J,CAChK,CAAC;AACJ,CAAC;AAED,SAAS,SAAS,CAAC,KAAiB;IAClC,MAAM,IAAI,GAAG,IAAI,QAAQ,CAAC,KAAK,CAAC,MAAM,EAAE,KAAK,CAAC,UAAU,EAAE,KAAK,CAAC,UAAU,CAAC,CAAC;IAC5E,IACE,IAAI,CAAC,SAAS,CAAC,CAAC,CAAC,KAAK,aAAa;QACnC,IAAI,CAAC,SAAS,CAAC,CAAC,CAAC,KAAK,aAAa,EACnC,CAAC;QACD,MAAM,IAAI,KAAK,CAAC,+CAA+C,CAAC,CAAC;IACnE,CAAC;IAED,qCAAqC;IACrC,IAAI,MAAM,GAAG,EAAE,CAAC;IAChB,IAAI,UAAU,GAAG,CAAC,CAAC;IACnB,IAAI,QAAQ,GAAG,CAAC,CAAC;IACjB,IAAI,aAAa,GAAG,CAAC,CAAC;IACtB,IAAI,WAAW,GAAG,CAAC,CAAC;IACpB,IAAI,SAAS,GAAG,CAAC,CAAC,CAAC;IACnB,IAAI,OAAO,GAAG,CAAC,CAAC;IAEhB,OAAO,MAAM,GAAG,CAAC,IAAI,KAAK,CAAC,UAAU,EAAE,CAAC;QACtC,MAAM,OAAO,GAAG,IAAI,CAAC,SAAS,CAAC,MAAM,CAAC,CAAC;QACvC,MAAM,SAAS,GAAG,IAAI,CAAC,SAAS,CAAC,MAAM,GAAG,CAAC,EAAE,IAAI,CAAC,CAAC;QACnD,IAAI,OAAO,KAAK,aAAa,EAAE,CAAC;YAC9B,WAAW,GAAG,IAAI,CAAC,SAAS,CAAC,MAAM,GAAG,CAAC,EAAE,IAAI,CAAC,CAAC;YAC/C,QAAQ,GAAG,IAAI,CAAC,SAAS,CAAC,MAAM,GAAG,EAAE,EAAE,IAAI,CAAC,CAAC;YAC7C,UAAU,GAAG,IAAI,CAAC,SAAS,CAAC,MAAM,GAAG,EAAE,EAAE,IAAI,CAAC,CAAC;YAC/C,aAAa,GAAG,IAAI,CAAC,SAAS,CAAC,MAAM,GAAG,EAAE,EAAE,IAAI,CAAC,CAAC;QACpD,CAAC;aAAM,IAAI,OAAO,KAAK,aAAa,EAAE,CAAC;YACrC,SAAS,GAAG,MAAM,GAAG,CAAC,CAAC;YACvB,OAAO,GAAG,SAAS,CAAC;YACpB,MAAM;QACR,CAAC;QACD,MAAM,IAAI,CAAC,GAAG,SAAS,GAAG,CAAC,SAAS,GAAG,CAAC,CAAC,CAAC;IAC5C,CAAC;IAED,IACE,SAAS,GAAG,CAAC;QACb,UAAU,KAAK,CAAC;QAChB,aAAa,KAAK,EAAE;QACpB,WAAW,KAAK,CAAC,EACjB,CAAC;QACD,MAAM,IAAI,KAAK,CACb,8EAA8E,WAAW,SAAS,aAAa,GAAG,CACnH,CAAC;IACJ,CAAC;IAED,MAAM,OAAO,GAAG,KAAK,CAAC,QAAQ,CAAC,SAAS,EAAE,SAAS,GAAG,OAAO,CAAC,CAAC;IAC/D,MAAM,WAAW,GAAG,eAAe,CAAC,OAAO,CAAC,CAAC;IAC7C,OAAO;QACL,GAAG,EAAE,aAAa,CAAC,WAAW,EAAE,QAAQ,IAAI,CAAC,CAAC;QAC9C,UAAU;QACV,QAAQ,EAAE,CAAC;KACZ,CAAC;AACJ,CAAC;AAED,gEAAgE;AAChE,SAAS,uBAAuB,CAC9B,KAAiB,EACjB,QAAgB,EAChB,MAAc;IAEd,IAAI,QAAQ,KAAK,MAAM,EAAE,CAAC;QACxB,OAAO,KAAK,CAAC;IACf,CAAC;IACD,MAAM,KAAK,GAAG,QAAQ,GAAG,MAAM,CAAC;IAChC,MAAM,MAAM,GAAG,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,MAAM,GAAG,KAAK,CAAC,CAAC;IAChD,MAAM,GAAG,GAAG,IAAI,UAAU,CAAC,MAAM,CAAC,CAAC;IACnC,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QAChC,MAAM,MAAM,GAAG,CAAC,GAAG,KAAK,CAAC;QACzB,MAAM,EAAE,GAAG,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC;QAC9B,MAAM,EAAE,GAAG,IAAI,CAAC,GAAG,CAAC,EAAE,GAAG,CAAC,EAAE,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;QAC9C,MAAM,IAAI,GAAG,MAAM,GAAG,EAAE,CAAC;QACzB,GAAG,CAAC,CAAC,CAAC,GAAG,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,EAAE,CAAC,GAAG,CAAC,CAAC,GAAG,IAAI,CAAC,GAAG,KAAK,CAAC,EAAE,CAAC,GAAG,IAAI,CAAC,CAAC;IACjE,CAAC;IACD,OAAO,GAAG,CAAC;AACb,CAAC;AAED,SAAS,YAAY,CAAC,EAAU,EAAE,UAAkB;IAClD,MAAM,OAAO,GAAG,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,GAAG,IAAI,CAAC,GAAG,UAAU,CAAC,CAAC;IACrD,OAAO,IAAI,UAAU,CAAC,OAAO,CAAC,CAAC;AACjC,CAAC;AAED,mDAAmD;AACnD,SAAS,QAAQ,CAAC,GAAe;IAC/B,IAAI,GAAG,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QACrB,OAAO,CAAC,CAAC;IACX,CAAC;IACD,IAAI,KAAK,GAAG,CAAC,CAAC;IACd,KAAK,MAAM,CAAC,IAAI,GAAG,EAAE,CAAC;QACpB,KAAK,IAAI,CAAC,GAAG,CAAC,CAAC;IACjB,CAAC;IACD,OAAO,IAAI,CAAC,IAAI,CAAC,KAAK,GAAG,GAAG,CAAC,MAAM,CAAC,CAAC;AACvC,CAAC;AAED,SAAS,UAAU,CAAC,KAAa;IAC/B,IAAI,KAAK,GAAG,SAAS,EAAE,CAAC;QACtB,OAAO,SAAS,CAAC;IACnB,CAAC;IACD,IAAI,KAAK,GAAG,SAAS,EAAE,CAAC;QACtB,OAAO,SAAS,CAAC;IACnB,CAAC;IACD,OAAO,KAAK,CAAC;AACf,CAAC;AAED,+DAA+D;AAC/D,SAAS,UAAU,CAAC,GAAe,EAAE,IAAY;IAC/C,MAAM,GAAG,GAAG,IAAI,UAAU,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC;IACvC,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,GAAG,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QACpC,GAAG,CAAC,CAAC,CAAC,GAAG,UAAU,CAAC,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,GAAG,IAAI,CAAC,CAAC,CAAC;IACjD,CAAC;IACD,OAAO,GAAG,CAAC;AACb,CAAC;AAED;;;;GAIG;AACH,MAAM,CAAC,MAAM,mBAAmB,GAAG,CAAC,EAAE,CAAC;AAEvC,wEAAwE;AACxE,MAAM,UAAU,cAAc,CAAC,IAAY;IACzC,OAAO,IAAI,CAAC,KAAK,CAAC,SAAS,GAAG,EAAE,IAAI,CAAC,IAAI,GAAG,EAAE,CAAC,CAAC,CAAC;AACnD,CAAC;AAED,MAAM,wBAAwB,GAAG,cAAc,CAAC,mBAAmB,CAAC,CAAC;AAErE;;;;;;;;;;;GAWG;AACH,MAAM,UAAU,YAAY,CAC1B,QAAiC,EACjC,kBAAkB,GAAG,wBAAwB;IAE7C,OAAO,QAAQ,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE;QACxB,MAAM,MAAM,GAAG,QAAQ,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC;QAC/B,IAAI,MAAM,KAAK,CAAC,EAAE,CAAC;YACjB,OAAO,EAAE,GAAG,CAAC,EAAE,CAAC;QAClB,CAAC;QACD,OAAO,EAAE,GAAG,CAAC,EAAE,GAAG,EAAE,UAAU,CAAC,CAAC,CAAC,GAAG,EAAE,kBAAkB,GAAG,MAAM,CAAC,EAAE,CAAC;IACvE,CAAC,CAAC,CAAC;AACL,CAAC;AAED;;;GAGG;AACH,MAAM,CAAC,KAAK,UAAU,cAAc,CAClC,QAAiC,EACjC,OAAoD;IAEpD,MAAM,EAAE,KAAK,EAAE,gBAAgB,EAAE,GAAG,OAAO,CAAC;IAE5C,MAAM,SAAS,GAAiB,EAAE,CAAC;IACnC,MAAM,GAAG,GAAG,YAAY,CAAC,KAAK,EAAE,gBAAgB,CAAC,CAAC;IAElD,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,QAAQ,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QACzC,MAAM,CAAC,GAAG,QAAQ,CAAC,CAAC,CAAC,CAAC;QACtB,SAAS,CAAC,IAAI,CACZ,uBAAuB,CAAC,CAAC,CAAC,GAAG,EAAE,CAAC,CAAC,UAAU,EAAE,gBAAgB,CAAC,CAC/D,CAAC;QACF,IAAI,CAAC,GAAG,QAAQ,CAAC,MAAM,GAAG,CAAC,IAAI,GAAG,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YAC9C,SAAS,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;QACtB,CAAC;IACH,CAAC;IAED,MAAM,YAAY,GAAG,SAAS,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,MAAM,EAAE,CAAC,CAAC,CAAC;IACjE,MAAM,MAAM,GAAG,IAAI,UAAU,CAAC,YAAY,CAAC,CAAC;IAC5C,IAAI,GAAG,GAAG,CAAC,CAAC;IACZ,KAAK,MAAM,CAAC,IAAI,SAAS,EAAE,CAAC;QAC1B,MAAM,CAAC,GAAG,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC;QACnB,GAAG,IAAI,CAAC,CAAC,MAAM,CAAC;IAClB,CAAC;IAED,MAAM,WAAW,GAAG,IAAI,UAAU,CAChC,MAAM,CAAC,MAAM,EACb,MAAM,CAAC,UAAU,EACjB,MAAM,CAAC,UAAU,CAClB,CAAC;IACF,OAAO,MAAM,aAAa,CAAC,WAAW,EAAE,gBAAgB,CAAC,CAAC;AAC5D,CAAC"}

package/dist/conversation/stitch.d.ts

@@ -0,0 +1,33 @@
+import type { ResolvedModel, Voice } from "../speech-provider.js";
+import type { ConversationTurn } from "./types.js";
+interface StitchInput<V extends Voice = Voice> {
+    readonly abortSignal?: AbortSignal;
+    readonly apiKey?: string;
+    readonly gapMs: number;
+    readonly headers?: Record<string, string>;
+    readonly maxConcurrency: number;
+    readonly maxRetries: number;
+    readonly normalizeVolume: boolean;
+    readonly resolvedPerTurn: readonly ResolvedModel<V>[];
+    readonly stitchOptionsPerTurn: readonly {
+        providerOptions: Record<string, unknown>;
+        mediaType: string;
+    }[];
+    readonly topLevelProviderOptions?: Record<string, unknown>;
+    readonly turns: readonly ConversationTurn<V>[];
+    readonly volumeDbfs?: number;
+}
+interface StitchOutput {
+    readonly audio: Uint8Array;
+    readonly mediaType: string;
+    readonly metadata: {
+        readonly inputChars: number;
+        readonly latencyMs: number;
+        readonly audioDurationMs?: number;
+    };
+    readonly providerMetadataPerTurn: readonly (Record<string, unknown> | undefined)[];
+    readonly warnings: readonly string[];
+}
+export declare function runStitch<V extends Voice>(input: StitchInput<V>): Promise<StitchOutput>;
+export {};
+//# sourceMappingURL=stitch.d.ts.map

package/dist/conversation/stitch.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"stitch.d.ts","sourceRoot":"","sources":["../../src/conversation/stitch.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAAE,aAAa,EAAE,KAAK,EAAE,MAAM,uBAAuB,CAAC;AAOlE,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,YAAY,CAAC;AAEnD,UAAU,WAAW,CAAC,CAAC,SAAS,KAAK,GAAG,KAAK;IAC3C,QAAQ,CAAC,WAAW,CAAC,EAAE,WAAW,CAAC;IACnC,QAAQ,CAAC,MAAM,CAAC,EAAE,MAAM,CAAC;IACzB,QAAQ,CAAC,KAAK,EAAE,MAAM,CAAC;IACvB,QAAQ,CAAC,OAAO,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC;IAC1C,QAAQ,CAAC,cAAc,EAAE,MAAM,CAAC;IAChC,QAAQ,CAAC,UAAU,EAAE,MAAM,CAAC;IAC5B,QAAQ,CAAC,eAAe,EAAE,OAAO,CAAC;IAClC,QAAQ,CAAC,eAAe,EAAE,SAAS,aAAa,CAAC,CAAC,CAAC,EAAE,CAAC;IACtD,QAAQ,CAAC,oBAAoB,EAAE,SAAS;QACtC,eAAe,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC;QACzC,SAAS,EAAE,MAAM,CAAC;KACnB,EAAE,CAAC;IACJ,QAAQ,CAAC,uBAAuB,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC;IAC3D,QAAQ,CAAC,KAAK,EAAE,SAAS,gBAAgB,CAAC,CAAC,CAAC,EAAE,CAAC;IAC/C,QAAQ,CAAC,UAAU,CAAC,EAAE,MAAM,CAAC;CAC9B;AAED,UAAU,YAAY;IACpB,QAAQ,CAAC,KAAK,EAAE,UAAU,CAAC;IAC3B,QAAQ,CAAC,SAAS,EAAE,MAAM,CAAC;IAC3B,QAAQ,CAAC,QAAQ,EAAE;QACjB,QAAQ,CAAC,UAAU,EAAE,MAAM,CAAC;QAC5B,QAAQ,CAAC,SAAS,EAAE,MAAM,CAAC;QAC3B,QAAQ,CAAC,eAAe,CAAC,EAAE,MAAM,CAAC;KACnC,CAAC;IACF,QAAQ,CAAC,uBAAuB,EAAE,SAAS,CACvC,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,GACvB,SAAS,CACZ,EAAE,CAAC;IACJ,QAAQ,CAAC,QAAQ,EAAE,SAAS,MAAM,EAAE,CAAC;CACtC;AA+BD,wBAAsB,SAAS,CAAC,CAAC,SAAS,KAAK,EAC7C,KAAK,EAAE,WAAW,CAAC,CAAC,CAAC,GACpB,OAAO,CAAC,YAAY,CAAC,CA+EvB"}

package/dist/conversation/stitch.js

@@ -0,0 +1,78 @@
+import { generateSpeech } from "../generate-speech.js";
+import { concatPcmToWav, dbfsToInt16Rms, decodeToPcm16, normalizeRms, } from "./pcm-concat.js";
+const TARGET_SAMPLE_RATE = 24_000;
+/**
+ * Run `worker(items[i], i)` for each item, capping in-flight executions at
+ * `concurrency`. Preserves input ordering in the returned array.
+ */
+async function mapWithConcurrency(items, concurrency, worker) {
+    const results = new Array(items.length);
+    let next = 0;
+    const runners = Array.from({ length: Math.min(Math.max(concurrency, 1), items.length) }, async () => {
+        while (true) {
+            const i = next++;
+            if (i >= items.length) {
+                return;
+            }
+            results[i] = await worker(items[i], i);
+        }
+    });
+    await Promise.all(runners);
+    return results;
+}
+export async function runStitch(input) {
+    const start = performance.now();
+    const perTurn = await mapWithConcurrency(input.turns, input.maxConcurrency, async (turn, i) => {
+        const resolved = input.resolvedPerTurn[i];
+        const stitchOpts = input.stitchOptionsPerTurn[i];
+        const mergedProviderOptions = {
+            ...input.topLevelProviderOptions,
+            ...turn.providerOptions,
+            ...stitchOpts.providerOptions,
+        };
+        const result = await generateSpeech({
+            model: resolved,
+            text: turn.text,
+            voice: turn.voice,
+            apiKey: input.apiKey,
+            providerOptions: mergedProviderOptions,
+            maxRetries: input.maxRetries,
+            abortSignal: input.abortSignal,
+            headers: input.headers,
+        });
+        // Prefer the mediaType from getStitchOptions over the response
+        // content-type: providers' response headers often omit the sample
+        // rate (e.g. Hume sends `audio/pcm` for what is actually 48 kHz),
+        // and getStitchOptions is the authoritative declaration of what
+        // the provider returns for the requested format.
+        const segment = decodeToPcm16(result.audio.uint8Array, stitchOpts.mediaType);
+        return { result, segment };
+    });
+    const segments = perTurn.map((p) => p.segment);
+    const leveledSegments = input.normalizeVolume
+        ? normalizeRms(segments, input.volumeDbfs == null ? undefined : dbfsToInt16Rms(input.volumeDbfs))
+        : segments;
+    const audio = await concatPcmToWav(leveledSegments, {
+        gapMs: input.gapMs,
+        targetSampleRate: TARGET_SAMPLE_RATE,
+    });
+    const totalSamples = perTurn.reduce((n, p) => n +
+        Math.round((p.segment.pcm.length / p.segment.sampleRate) * TARGET_SAMPLE_RATE), 0) +
+        (perTurn.length - 1) *
+            Math.round((input.gapMs / 1000) * TARGET_SAMPLE_RATE);
+    const audioDurationMs = Math.round((totalSamples / TARGET_SAMPLE_RATE) * 1000);
+    const warnings = perTurn.flatMap((p) => p.result.warnings ?? []);
+    const providerMetadataPerTurn = perTurn.map((p) => p.result.providerMetadata);
+    return {
+        audio,
+        mediaType: "audio/wav",
+        metadata: {
+            inputChars: input.turns.reduce((n, t) => n + t.text.length, 0),
+            latencyMs: Math.round(performance.now() - start),
+            audioDurationMs,
+        },
+        providerMetadataPerTurn,
+        warnings,
+    };
+}
+//# sourceMappingURL=stitch.js.map

package/dist/conversation/stitch.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"stitch.js","sourceRoot":"","sources":["../../src/conversation/stitch.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,cAAc,EAAE,MAAM,uBAAuB,CAAC;AAEvD,OAAO,EACL,cAAc,EACd,cAAc,EACd,aAAa,EACb,YAAY,GACb,MAAM,iBAAiB,CAAC;AAoCzB,MAAM,kBAAkB,GAAG,MAAM,CAAC;AAElC;;;GAGG;AACH,KAAK,UAAU,kBAAkB,CAC/B,KAAmB,EACnB,WAAmB,EACnB,MAA8C;IAE9C,MAAM,OAAO,GAAQ,IAAI,KAAK,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC;IAC7C,IAAI,IAAI,GAAG,CAAC,CAAC;IACb,MAAM,OAAO,GAAG,KAAK,CAAC,IAAI,CACxB,EAAE,MAAM,EAAE,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,WAAW,EAAE,CAAC,CAAC,EAAE,KAAK,CAAC,MAAM,CAAC,EAAE,EAC5D,KAAK,IAAI,EAAE;QACT,OAAO,IAAI,EAAE,CAAC;YACZ,MAAM,CAAC,GAAG,IAAI,EAAE,CAAC;YACjB,IAAI,CAAC,IAAI,KAAK,CAAC,MAAM,EAAE,CAAC;gBACtB,OAAO;YACT,CAAC;YACD,OAAO,CAAC,CAAC,CAAC,GAAG,MAAM,MAAM,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC;QACzC,CAAC;IACH,CAAC,CACF,CAAC;IACF,MAAM,OAAO,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC;IAC3B,OAAO,OAAO,CAAC;AACjB,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,SAAS,CAC7B,KAAqB;IAErB,MAAM,KAAK,GAAG,WAAW,CAAC,GAAG,EAAE,CAAC;IAEhC,MAAM,OAAO,GAAG,MAAM,kBAAkB,CACtC,KAAK,CAAC,KAAK,EACX,KAAK,CAAC,cAAc,EACpB,KAAK,EAAE,IAAI,EAAE,CAAC,EAAE,EAAE;QAChB,MAAM,QAAQ,GAAG,KAAK,CAAC,eAAe,CAAC,CAAC,CAAC,CAAC;QAC1C,MAAM,UAAU,GAAG,KAAK,CAAC,oBAAoB,CAAC,CAAC,CAAC,CAAC;QACjD,MAAM,qBAAqB,GAAG;YAC5B,GAAG,KAAK,CAAC,uBAAuB;YAChC,GAAG,IAAI,CAAC,eAAe;YACvB,GAAG,UAAU,CAAC,eAAe;SAC9B,CAAC;QACF,MAAM,MAAM,GAAG,MAAM,cAAc,CAAC;YAClC,KAAK,EAAE,QAAQ;YACf,IAAI,EAAE,IAAI,CAAC,IAAI;YACf,KAAK,EAAE,IAAI,CAAC,KAAK;YACjB,MAAM,EAAE,KAAK,CAAC,MAAM;YACpB,eAAe,EAAE,qBAAqB;YACtC,UAAU,EAAE,KAAK,CAAC,UAAU;YAC5B,WAAW,EAAE,KAAK,CAAC,WAAW;YAC9B,OAAO,EAAE,KAAK,CAAC,OAAO;SACvB,CAAC,CAAC;QACH,+DAA+D;QAC/D,kEAAkE;QAClE,kEAAkE;QAClE,gEAAgE;QAChE,iDAAiD;QACjD,MAAM,OAAO,GAAG,aAAa,CAC3B,MAAM,CAAC,KAAK,CAAC,UAAU,EACvB,UAAU,CAAC,SAAS,CACrB,CAAC;QACF,OAAO,EAAE,MAAM,EAAE,OAAO,EAAE,CAAC;IAC7B,CAAC,CACF,CAAC;IAEF,MAAM,QAAQ,GAAG,OAAO,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC;IAC/C,MAAM,eAAe,GAAG,KAAK,CAAC,eAAe;QAC3C,CAAC,CAAC,YAAY,CACV,QAAQ,EACR,KAAK,CAAC,UAAU,IAAI,IAAI,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,cAAc,CAAC,KAAK,CAAC,UAAU,CAAC,CACxE;QACH,CAAC,CAAC,QAAQ,CAAC;IAEb,MAAM,KAAK,GAAG,MAAM,cAAc,CAAC,eAAe,EAAE;QAClD,KAAK,EAAE,KAAK,CAAC,KAAK;QAClB,gBAAgB,EAAE,kBAAkB;KACrC,CAAC,CAAC;IAEH,MAAM,YAAY,GAChB,OAAO,CAAC,MAAM,CACZ,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CACP,CAAC;QACD,IAAI,CAAC,KAAK,CACR,CAAC,CAAC,CAAC,OAAO,CAAC,GAAG,CAAC,MAAM,GAAG,CAAC,CAAC,OAAO,CAAC,UAAU,CAAC,GAAG,kBAAkB,CACnE,EACH,CAAC,CACF;QACD,CAAC,OAAO,CAAC,MAAM,GAAG,CAAC,CAAC;YAClB,IAAI,CAAC,KAAK,CAAC,CAAC,KAAK,CAAC,KAAK,GAAG,IAAI,CAAC,GAAG,kBAAkB,CAAC,CAAC;IAC1D,MAAM,eAAe,GAAG,IAAI,CAAC,KAAK,CAChC,CAAC,YAAY,GAAG,kBAAkB,CAAC,GAAG,IAAI,CAC3C,CAAC;IAEF,MAAM,QAAQ,GAAG,OAAO,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,MAAM,CAAC,QAAQ,IAAI,EAAE,CAAC,CAAC;IACjE,MAAM,uBAAuB,GAAG,OAAO,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,MAAM,CAAC,gBAAgB,CAAC,CAAC;IAE9E,OAAO;QACL,KAAK;QACL,SAAS,EAAE,WAAW;QACtB,QAAQ,EAAE;YACR,UAAU,EAAE,KAAK,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,IAAI,CAAC,MAAM,EAAE,CAAC,CAAC;YAC9D,SAAS,EAAE,IAAI,CAAC,KAAK,CAAC,WAAW,CAAC,GAAG,EAAE,GAAG,KAAK,CAAC;YAChD,eAAe;SAChB;QACD,uBAAuB;QACvB,QAAQ;KACT,CAAC;AACJ,CAAC"}

package/dist/conversation/types.d.ts

@@ -0,0 +1,39 @@
+import type { ResolvedModel, Voice } from "../speech-provider.js";
+export interface ConversationTurn<V extends Voice = Voice> {
+    readonly model?: string | ResolvedModel<V>;
+    readonly providerOptions?: Record<string, unknown>;
+    readonly text: string;
+    readonly voice: V;
+}
+export interface GenerateConversationOptions<V extends Voice = Voice> {
+    readonly abortSignal?: AbortSignal;
+    readonly apiKey?: string;
+    readonly gapMs?: number;
+    readonly headers?: Record<string, string>;
+    readonly maxConcurrency?: number;
+    readonly maxRetries?: number;
+    readonly model?: string | ResolvedModel<V>;
+    /**
+     * RMS-normalize the output audio to an absolute target level (see
+     * `volumeDbfs` for the level itself, default -20 dBFS — the broadcast /
+     * podcast voice standard). Every call to generateConversation produces
+     * output at the same loudness regardless of which providers or content
+     * are used, so two separate conversations can be played back-to-back
+     * without the listener adjusting volume. Roughly two O(N) passes over
+     * the int16 PCM samples — cheap. Pass `false` to skip the step entirely
+     * (~zero work) and keep the raw provider levels. Applied on both the
+     * stitch and native dialogue paths, provided the chosen provider
+     * exposes a decodable PCM/WAV mode via `getStitchOptions`. Default: true.
+     */
+    readonly normalizeVolume?: boolean;
+    readonly providerOptions?: Record<string, unknown>;
+    readonly turns: readonly ConversationTurn<V>[];
+    /**
+     * Target loudness in dBFS for `normalizeVolume`. Must be ≤ 0 (0 dBFS is
+     * the int16 ceiling). Lower values are quieter — -20 leaves ~20 dB of
+     * peak headroom so typical TTS speech doesn't clip after gain. Ignored
+     * when `normalizeVolume` is `false`. Default: -20.
+     */
+    readonly volumeDbfs?: number;
+}
+//# sourceMappingURL=types.d.ts.map

package/dist/conversation/types.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"types.d.ts","sourceRoot":"","sources":["../../src/conversation/types.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,aAAa,EAAE,KAAK,EAAE,MAAM,uBAAuB,CAAC;AAElE,MAAM,WAAW,gBAAgB,CAAC,CAAC,SAAS,KAAK,GAAG,KAAK;IACvD,QAAQ,CAAC,KAAK,CAAC,EAAE,MAAM,GAAG,aAAa,CAAC,CAAC,CAAC,CAAC;IAC3C,QAAQ,CAAC,eAAe,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC;IACnD,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAC;IACtB,QAAQ,CAAC,KAAK,EAAE,CAAC,CAAC;CACnB;AAED,MAAM,WAAW,2BAA2B,CAAC,CAAC,SAAS,KAAK,GAAG,KAAK;IAClE,QAAQ,CAAC,WAAW,CAAC,EAAE,WAAW,CAAC;IACnC,QAAQ,CAAC,MAAM,CAAC,EAAE,MAAM,CAAC;IACzB,QAAQ,CAAC,KAAK,CAAC,EAAE,MAAM,CAAC;IACxB,QAAQ,CAAC,OAAO,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC;IAC1C,QAAQ,CAAC,cAAc,CAAC,EAAE,MAAM,CAAC;IACjC,QAAQ,CAAC,UAAU,CAAC,EAAE,MAAM,CAAC;IAC7B,QAAQ,CAAC,KAAK,CAAC,EAAE,MAAM,GAAG,aAAa,CAAC,CAAC,CAAC,CAAC;IAC3C;;;;;;;;;;;OAWG;IACH,QAAQ,CAAC,eAAe,CAAC,EAAE,OAAO,CAAC;IACnC,QAAQ,CAAC,eAAe,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC;IACnD,QAAQ,CAAC,KAAK,EAAE,SAAS,gBAAgB,CAAC,CAAC,CAAC,EAAE,CAAC;IAC/C;;;;;OAKG;IACH,QAAQ,CAAC,UAAU,CAAC,EAAE,MAAM,CAAC;CAC9B"}

package/dist/conversation/types.js

@@ -0,0 +1,2 @@
+export {};
+//# sourceMappingURL=types.js.map

package/dist/conversation/types.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"types.js","sourceRoot":"","sources":["../../src/conversation/types.ts"],"names":[],"mappings":""}

package/dist/conversation/validate.d.ts

@@ -0,0 +1,19 @@
+import type { ConversationTurn, GenerateConversationOptions } from "./types.js";
+/**
+ * Stable key for a voice so we can count unique voices across turns within
+ * one call. String voices and URL voices use their value; binary
+ * `Uint8Array` audio voices use object-reference identity (two distinct
+ * buffers with the same length/endpoints would otherwise collide).
+ */
+export declare function voiceKey(voice: ConversationTurn["voice"], refIds: WeakMap<object, number>, refCounter: {
+    next: number;
+}): string;
+/** Build a fresh ref-id context for a single conversation. */
+export declare function newVoiceKeyContext(): {
+    refIds: WeakMap<object, number>;
+    refCounter: {
+        next: number;
+    };
+};
+export declare function validateConversationInput(options: GenerateConversationOptions): void;
+//# sourceMappingURL=validate.d.ts.map

package/dist/conversation/validate.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"validate.d.ts","sourceRoot":"","sources":["../../src/conversation/validate.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAAE,gBAAgB,EAAE,2BAA2B,EAAE,MAAM,YAAY,CAAC;AAIhF;;;;;GAKG;AACH,wBAAgB,QAAQ,CACtB,KAAK,EAAE,gBAAgB,CAAC,OAAO,CAAC,EAChC,MAAM,EAAE,OAAO,CAAC,MAAM,EAAE,MAAM,CAAC,EAC/B,UAAU,EAAE;IAAE,IAAI,EAAE,MAAM,CAAA;CAAE,GAC3B,MAAM,CAiBR;AAED,8DAA8D;AAC9D,wBAAgB,kBAAkB,IAAI;IACpC,MAAM,EAAE,OAAO,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC;IAChC,UAAU,EAAE;QAAE,IAAI,EAAE,MAAM,CAAA;KAAE,CAAC;CAC9B,CAEA;AAED,wBAAgB,yBAAyB,CACvC,OAAO,EAAE,2BAA2B,GACnC,IAAI,CA4BN"}

package/dist/conversation/validate.js

@@ -0,0 +1,50 @@
+import { ConversationInputError } from "./errors.js";
+const MAX_UNIQUE_VOICES = 4;
+/**
+ * Stable key for a voice so we can count unique voices across turns within
+ * one call. String voices and URL voices use their value; binary
+ * `Uint8Array` audio voices use object-reference identity (two distinct
+ * buffers with the same length/endpoints would otherwise collide).
+ */
+export function voiceKey(voice, refIds, refCounter) {
+    if (typeof voice === "string") {
+        return `s:${voice}`;
+    }
+    if ("url" in voice) {
+        return `u:${voice.url}`;
+    }
+    if ("audio" in voice && typeof voice.audio === "string") {
+        return `a:${voice.audio}`;
+    }
+    // Object voice (Uint8Array audio, or any other shape) — key by reference.
+    let id = refIds.get(voice);
+    if (id === undefined) {
+        id = refCounter.next++;
+        refIds.set(voice, id);
+    }
+    return `o:${id}`;
+}
+/** Build a fresh ref-id context for a single conversation. */
+export function newVoiceKeyContext() {
+    return { refIds: new WeakMap(), refCounter: { next: 0 } };
+}
+export function validateConversationInput(options) {
+    if (options.turns.length === 0) {
+        throw new ConversationInputError("generateConversation requires at least one turn.");
+    }
+    for (let i = 0; i < options.turns.length; i++) {
+        const turn = options.turns[i];
+        if (turn.text.trim().length === 0) {
+            throw new ConversationInputError(`turns[${i}].text must not be empty.`);
+        }
+        if (options.model == null && turn.model == null) {
+            throw new ConversationInputError(`turns[${i}]: model must be set, either at top-level or on the turn.`);
+        }
+    }
+    const ctx = newVoiceKeyContext();
+    const uniqueVoices = new Set(options.turns.map((t) => voiceKey(t.voice, ctx.refIds, ctx.refCounter)));
+    if (uniqueVoices.size > MAX_UNIQUE_VOICES) {
+        throw new ConversationInputError(`generateConversation accepts at most 4 unique voices; got ${uniqueVoices.size}.`);
+    }
+}
+//# sourceMappingURL=validate.js.map

package/dist/conversation/validate.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"validate.js","sourceRoot":"","sources":["../../src/conversation/validate.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,sBAAsB,EAAE,MAAM,aAAa,CAAC;AAGrD,MAAM,iBAAiB,GAAG,CAAC,CAAC;AAE5B;;;;;GAKG;AACH,MAAM,UAAU,QAAQ,CACtB,KAAgC,EAChC,MAA+B,EAC/B,UAA4B;IAE5B,IAAI,OAAO,KAAK,KAAK,QAAQ,EAAE,CAAC;QAC9B,OAAO,KAAK,KAAK,EAAE,CAAC;IACtB,CAAC;IACD,IAAI,KAAK,IAAI,KAAK,EAAE,CAAC;QACnB,OAAO,KAAK,KAAK,CAAC,GAAG,EAAE,CAAC;IAC1B,CAAC;IACD,IAAI,OAAO,IAAI,KAAK,IAAI,OAAO,KAAK,CAAC,KAAK,KAAK,QAAQ,EAAE,CAAC;QACxD,OAAO,KAAK,KAAK,CAAC,KAAK,EAAE,CAAC;IAC5B,CAAC;IACD,0EAA0E;IAC1E,IAAI,EAAE,GAAG,MAAM,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC;IAC3B,IAAI,EAAE,KAAK,SAAS,EAAE,CAAC;QACrB,EAAE,GAAG,UAAU,CAAC,IAAI,EAAE,CAAC;QACvB,MAAM,CAAC,GAAG,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC;IACxB,CAAC;IACD,OAAO,KAAK,EAAE,EAAE,CAAC;AACnB,CAAC;AAED,8DAA8D;AAC9D,MAAM,UAAU,kBAAkB;IAIhC,OAAO,EAAE,MAAM,EAAE,IAAI,OAAO,EAAE,EAAE,UAAU,EAAE,EAAE,IAAI,EAAE,CAAC,EAAE,EAAE,CAAC;AAC5D,CAAC;AAED,MAAM,UAAU,yBAAyB,CACvC,OAAoC;IAEpC,IAAI,OAAO,CAAC,KAAK,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QAC/B,MAAM,IAAI,sBAAsB,CAC9B,kDAAkD,CACnD,CAAC;IACJ,CAAC;IAED,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,OAAO,CAAC,KAAK,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QAC9C,MAAM,IAAI,GAAG,OAAO,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC;QAC9B,IAAI,IAAI,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YAClC,MAAM,IAAI,sBAAsB,CAAC,SAAS,CAAC,2BAA2B,CAAC,CAAC;QAC1E,CAAC;QACD,IAAI,OAAO,CAAC,KAAK,IAAI,IAAI,IAAI,IAAI,CAAC,KAAK,IAAI,IAAI,EAAE,CAAC;YAChD,MAAM,IAAI,sBAAsB,CAC9B,SAAS,CAAC,2DAA2D,CACtE,CAAC;QACJ,CAAC;IACH,CAAC;IAED,MAAM,GAAG,GAAG,kBAAkB,EAAE,CAAC;IACjC,MAAM,YAAY,GAAG,IAAI,GAAG,CAC1B,OAAO,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,QAAQ,CAAC,CAAC,CAAC,KAAK,EAAE,GAAG,CAAC,MAAM,EAAE,GAAG,CAAC,UAAU,CAAC,CAAC,CACxE,CAAC;IACF,IAAI,YAAY,CAAC,IAAI,GAAG,iBAAiB,EAAE,CAAC;QAC1C,MAAM,IAAI,sBAAsB,CAC9B,6DAA6D,YAAY,CAAC,IAAI,GAAG,CAClF,CAAC;IACJ,CAAC;AACH,CAAC"}

package/dist/errors.d.ts

@@ -20,4 +20,7 @@ export declare class NoSpeechGeneratedError extends SpeechSDKError {
 export declare class StreamingNotSupportedError extends SpeechSDKError {
     constructor(model: string);
 }
+export declare class VolumeAdjustmentUnsupportedError extends SpeechSDKError {
+    constructor(model: string);
+}
 //# sourceMappingURL=errors.d.ts.map

package/dist/errors.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"errors.d.ts","sourceRoot":"","sources":["../src/errors.ts"],"names":[],"mappings":"AAAA,qBAAa,cAAe,SAAQ,KAAK;gBAC3B,OAAO,EAAE,MAAM,EAAE,OAAO,CAAC,EAAE;QAAE,KAAK,CAAC,EAAE,OAAO,CAAA;KAAE;CAI3D;AAED,qBAAa,QAAS,SAAQ,cAAc;IAC1C,QAAQ,CAAC,UAAU,EAAE,MAAM,CAAC;IAC5B,QAAQ,CAAC,YAAY,CAAC,EAAE,OAAO,CAAC;IAChC,QAAQ,CAAC,KAAK,EAAE,MAAM,CAAC;gBAGrB,OAAO,EAAE,MAAM,EACf,OAAO,EAAE;QACP,UAAU,EAAE,MAAM,CAAC;QACnB,KAAK,EAAE,MAAM,CAAC;QACd,YAAY,CAAC,EAAE,OAAO,CAAC;QACvB,KAAK,CAAC,EAAE,OAAO,CAAC;KACjB;CAQJ;AAED,qBAAa,sBAAuB,SAAQ,cAAc;gBAC5C,OAAO,CAAC,EAAE,MAAM;CAI7B;AAED,qBAAa,0BAA2B,SAAQ,cAAc;gBAChD,KAAK,EAAE,MAAM;CAM1B"}
+{"version":3,"file":"errors.d.ts","sourceRoot":"","sources":["../src/errors.ts"],"names":[],"mappings":"AAAA,qBAAa,cAAe,SAAQ,KAAK;gBAC3B,OAAO,EAAE,MAAM,EAAE,OAAO,CAAC,EAAE;QAAE,KAAK,CAAC,EAAE,OAAO,CAAA;KAAE;CAI3D;AAED,qBAAa,QAAS,SAAQ,cAAc;IAC1C,QAAQ,CAAC,UAAU,EAAE,MAAM,CAAC;IAC5B,QAAQ,CAAC,YAAY,CAAC,EAAE,OAAO,CAAC;IAChC,QAAQ,CAAC,KAAK,EAAE,MAAM,CAAC;gBAGrB,OAAO,EAAE,MAAM,EACf,OAAO,EAAE;QACP,UAAU,EAAE,MAAM,CAAC;QACnB,KAAK,EAAE,MAAM,CAAC;QACd,YAAY,CAAC,EAAE,OAAO,CAAC;QACvB,KAAK,CAAC,EAAE,OAAO,CAAC;KACjB;CAQJ;AAED,qBAAa,sBAAuB,SAAQ,cAAc;gBAC5C,OAAO,CAAC,EAAE,MAAM;CAI7B;AAED,qBAAa,0BAA2B,SAAQ,cAAc;gBAChD,KAAK,EAAE,MAAM;CAM1B;AAED,qBAAa,gCAAiC,SAAQ,cAAc;gBACtD,KAAK,EAAE,MAAM;CAM1B"}

package/dist/errors.js

@@ -28,4 +28,10 @@ export class StreamingNotSupportedError extends SpeechSDKError {
         this.name = "StreamingNotSupportedError";
     }
 }
+export class VolumeAdjustmentUnsupportedError extends SpeechSDKError {
+    constructor(model) {
+        super(`volumeDbfs is not supported by ${model}: the provider doesn't expose a decodable PCM/WAV output mode.`);
+        this.name = "VolumeAdjustmentUnsupportedError";
+    }
+}
 //# sourceMappingURL=errors.js.map

package/dist/errors.js.map

@@ -1 +1 @@
-{"version":3,"file":"errors.js","sourceRoot":"","sources":["../src/errors.ts"],"names":[],"mappings":"AAAA,MAAM,OAAO,cAAe,SAAQ,KAAK;IACvC,YAAY,OAAe,EAAE,OAA6B;QACxD,KAAK,CAAC,OAAO,EAAE,OAAO,CAAC,CAAC;QACxB,IAAI,CAAC,IAAI,GAAG,gBAAgB,CAAC;IAC/B,CAAC;CACF;AAED,MAAM,OAAO,QAAS,SAAQ,cAAc;IACjC,UAAU,CAAS;IACnB,YAAY,CAAW;IACvB,KAAK,CAAS;IAEvB,YACE,OAAe,EACf,OAKC;QAED,KAAK,CAAC,OAAO,EAAE,EAAE,KAAK,EAAE,OAAO,CAAC,KAAK,EAAE,CAAC,CAAC;QACzC,IAAI,CAAC,IAAI,GAAG,UAAU,CAAC;QACvB,IAAI,CAAC,UAAU,GAAG,OAAO,CAAC,UAAU,CAAC;QACrC,IAAI,CAAC,KAAK,GAAG,OAAO,CAAC,KAAK,CAAC;QAC3B,IAAI,CAAC,YAAY,GAAG,OAAO,CAAC,YAAY,CAAC;IAC3C,CAAC;CACF;AAED,MAAM,OAAO,sBAAuB,SAAQ,cAAc;IACxD,YAAY,OAAgB;QAC1B,KAAK,CAAC,OAAO,IAAI,gCAAgC,CAAC,CAAC;QACnD,IAAI,CAAC,IAAI,GAAG,wBAAwB,CAAC;IACvC,CAAC;CACF;AAED,MAAM,OAAO,0BAA2B,SAAQ,cAAc;IAC5D,YAAY,KAAa;QACvB,KAAK,CACH,iCAAiC,KAAK,iCAAiC,CACxE,CAAC;QACF,IAAI,CAAC,IAAI,GAAG,4BAA4B,CAAC;IAC3C,CAAC;CACF"}
+{"version":3,"file":"errors.js","sourceRoot":"","sources":["../src/errors.ts"],"names":[],"mappings":"AAAA,MAAM,OAAO,cAAe,SAAQ,KAAK;IACvC,YAAY,OAAe,EAAE,OAA6B;QACxD,KAAK,CAAC,OAAO,EAAE,OAAO,CAAC,CAAC;QACxB,IAAI,CAAC,IAAI,GAAG,gBAAgB,CAAC;IAC/B,CAAC;CACF;AAED,MAAM,OAAO,QAAS,SAAQ,cAAc;IACjC,UAAU,CAAS;IACnB,YAAY,CAAW;IACvB,KAAK,CAAS;IAEvB,YACE,OAAe,EACf,OAKC;QAED,KAAK,CAAC,OAAO,EAAE,EAAE,KAAK,EAAE,OAAO,CAAC,KAAK,EAAE,CAAC,CAAC;QACzC,IAAI,CAAC,IAAI,GAAG,UAAU,CAAC;QACvB,IAAI,CAAC,UAAU,GAAG,OAAO,CAAC,UAAU,CAAC;QACrC,IAAI,CAAC,KAAK,GAAG,OAAO,CAAC,KAAK,CAAC;QAC3B,IAAI,CAAC,YAAY,GAAG,OAAO,CAAC,YAAY,CAAC;IAC3C,CAAC;CACF;AAED,MAAM,OAAO,sBAAuB,SAAQ,cAAc;IACxD,YAAY,OAAgB;QAC1B,KAAK,CAAC,OAAO,IAAI,gCAAgC,CAAC,CAAC;QACnD,IAAI,CAAC,IAAI,GAAG,wBAAwB,CAAC;IACvC,CAAC;CACF;AAED,MAAM,OAAO,0BAA2B,SAAQ,cAAc;IAC5D,YAAY,KAAa;QACvB,KAAK,CACH,iCAAiC,KAAK,iCAAiC,CACxE,CAAC;QACF,IAAI,CAAC,IAAI,GAAG,4BAA4B,CAAC;IAC3C,CAAC;CACF;AAED,MAAM,OAAO,gCAAiC,SAAQ,cAAc;IAClE,YAAY,KAAa;QACvB,KAAK,CACH,kCAAkC,KAAK,gEAAgE,CACxG,CAAC;QACF,IAAI,CAAC,IAAI,GAAG,kCAAkC,CAAC;IACjD,CAAC;CACF"}

package/dist/generate-conversation.d.ts

@@ -0,0 +1,6 @@
+import type { GenerateConversationOptions } from "./conversation/types.js";
+import type { Voice } from "./speech-provider.js";
+import type { SpeechResult } from "./speech-result.js";
+export type { ConversationTurn, GenerateConversationOptions, } from "./conversation/types.js";
+export declare function generateConversation<V extends Voice = Voice>(options: GenerateConversationOptions<V>): Promise<SpeechResult>;
+//# sourceMappingURL=generate-conversation.d.ts.map

package/dist/generate-conversation.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"generate-conversation.d.ts","sourceRoot":"","sources":["../src/generate-conversation.ts"],"names":[],"mappings":"AAGA,OAAO,KAAK,EAAE,2BAA2B,EAAE,MAAM,yBAAyB,CAAC;AAK3E,OAAO,KAAK,EAAiB,KAAK,EAAE,MAAM,sBAAsB,CAAC;AACjE,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,oBAAoB,CAAC;AAGvD,YAAY,EACV,gBAAgB,EAChB,2BAA2B,GAC5B,MAAM,yBAAyB,CAAC;AAMjC,wBAAsB,oBAAoB,CAAC,CAAC,SAAS,KAAK,GAAG,KAAK,EAChE,OAAO,EAAE,2BAA2B,CAAC,CAAC,CAAC,GACtC,OAAO,CAAC,YAAY,CAAC,CAuEvB"}