@qvac/sdk 0.11.0 → 0.12.0

package/dist/examples/transcription/parakeet-sortformer.js

@@ -1,75 +1,78 @@
 /**
- * Parakeet Sortformer diarization + TDT transcription example.
+ * Parakeet Sortformer diarization + TDT transcription pipeline.
  *
  * Usage:
- *   bun examples/transcription/parakeet-sortformer.ts [sortformer-src] [path-to-audio]
+ *   bun run examples/transcription/parakeet-sortformer.ts [sortformer-gguf] [wav-file]
  *
- * This example requires a test audio file (default: examples/audio/diarization-sample-16k.wav).
- * Sample audio files are available in the QVAC source repository, but not included in the published npm package.
- * Pass a custom audio path as the second argument, or download the default audio into examples/audio/:
+ * Two-step flow: Sortformer v2.1 diarizes the audio, then TDT transcribes each
+ * speaker segment. Defaults to registry GGUFs and
+ * `examples/audio/diarization-sample-16k.wav`. For live streaming + AOSC, see
+ * `parakeet-sortformer-streaming.ts`.
+ *
+ * Sample audio is in the QVAC source repo but not the published npm package.
+ * Download the default file into `examples/audio/`:
  *   https://github.com/tetherto/qvac/blob/main/packages/sdk/examples/audio/diarization-sample-16k.wav
  */
-import { loadModel, unloadModel, transcribe, PARAKEET_TDT_ENCODER_FP32, PARAKEET_TDT_DECODER_FP32, PARAKEET_TDT_VOCAB, PARAKEET_TDT_PREPROCESSOR_FP32, PARAKEET_SORTFORMER_FP32, } from "@qvac/sdk";
+import { loadModel, unloadModel, transcribe, PARAKEET_TDT_0_6B_V3_Q8_0, PARAKEET_SORTFORMER_4SPK_V2_1_Q8_0, } from "@qvac/sdk";
 import { dirname, join } from "path";
 import { fileURLToPath } from "url";
 import { readFileSync, writeFileSync, mkdirSync } from "fs";
 import { tmpdir } from "os";
 const __dirname = dirname(fileURLToPath(import.meta.url));
 const args = process.argv.slice(2);
-const sortformerSrc = args[0] ?? PARAKEET_SORTFORMER_FP32;
+const sortformerSrc = args[0] ?? PARAKEET_SORTFORMER_4SPK_V2_1_Q8_0;
 const defaultAudioPath = join(__dirname, "..", "audio", "diarization-sample-16k.wav");
 const audioFilePath = args[1] ?? defaultAudioPath;
-// ── Step 1: Diarize with Sortformer ──
-const sfModelId = await loadModel({
-    modelSrc: sortformerSrc,
-    modelType: "parakeet",
-    modelConfig: {
-        modelType: "sortformer",
-        parakeetSortformerSrc: sortformerSrc,
-    },
-});
-const diarization = await transcribe({
-    modelId: sfModelId,
-    audioChunk: audioFilePath,
-});
-await unloadModel({ modelId: sfModelId });
-const segments = parseDiarization(diarization);
-// ── Step 2: Transcribe each segment with TDT ──
-const tdtModelId = await loadModel({
-    modelSrc: PARAKEET_TDT_ENCODER_FP32,
-    modelType: "parakeet",
-    modelConfig: {
-        parakeetEncoderSrc: PARAKEET_TDT_ENCODER_FP32,
-        parakeetDecoderSrc: PARAKEET_TDT_DECODER_FP32,
-        parakeetVocabSrc: PARAKEET_TDT_VOCAB,
-        parakeetPreprocessorSrc: PARAKEET_TDT_PREPROCESSOR_FP32,
-    },
-});
-const pcm = readPcm(audioFilePath);
-const sliceDir = join(tmpdir(), `qvac-diarize-${Date.now()}`);
-mkdirSync(sliceDir, { recursive: true });
-const results = [];
-for (let i = 0; i < segments.length; i++) {
-    const seg = segments[i];
-    const slicePath = join(sliceDir, `seg-${i}.wav`);
-    if (!writeWavSlice(pcm, seg.start, seg.end, slicePath)) {
-        results.push({ ...seg, text: "[No speech detected]" });
-        continue;
+try {
+    // ── Step 1: Diarize with Sortformer ──
+    const sfModelId = await loadModel({
+        modelSrc: sortformerSrc,
+        modelType: "parakeet",
+    });
+    const diarization = await transcribe({
+        modelId: sfModelId,
+        audioChunk: audioFilePath,
+    });
+    await unloadModel({ modelId: sfModelId });
+    const segments = parseDiarization(diarization);
+    // ── Step 2: Transcribe each segment with TDT ──
+    const tdtModelId = await loadModel({
+        modelSrc: PARAKEET_TDT_0_6B_V3_Q8_0,
+        modelType: "parakeet",
+    });
+    const pcm = readPcm(audioFilePath);
+    const sliceDir = join(tmpdir(), `qvac-diarize-${Date.now()}`);
+    mkdirSync(sliceDir, { recursive: true });
+    const results = [];
+    for (let i = 0; i < segments.length; i++) {
+        const seg = segments[i];
+        const slicePath = join(sliceDir, `seg-${i}.wav`);
+        if (!writeWavSlice(pcm, seg.start, seg.end, slicePath)) {
+            results.push({ ...seg, text: "[No speech detected]" });
+            continue;
+        }
+        const text = await transcribe({
+            modelId: tdtModelId,
+            audioChunk: slicePath,
+        });
+        results.push({ ...seg, text: text.trim() || "[No speech detected]" });
+    }
+    await unloadModel({ modelId: tdtModelId });
+    // ── Step 3: Merge consecutive same-speaker segments and print ──
+    const merged = mergeSpeakers(results);
+    console.log("\n=== DIARIZED TRANSCRIPTION ===");
+    console.log("=".repeat(60));
+    for (const entry of merged) {
+        console.log(`Speaker ${entry.speaker} (${entry.start.toFixed(2)}s - ${entry.end.toFixed(2)}s):`);
+        console.log(`  ${entry.text}\n`);
     }
-    const text = await transcribe({ modelId: tdtModelId, audioChunk: slicePath });
-    results.push({ ...seg, text: text.trim() || "[No speech detected]" });
+    console.log("=".repeat(60));
+    console.log("\nDone!");
 }
-await unloadModel({ modelId: tdtModelId });
-// ── Step 3: Merge consecutive same-speaker segments and print ──
-const merged = mergeSpeakers(results);
-console.log("\n=== DIARIZED TRANSCRIPTION ===");
-console.log("=".repeat(60));
-for (const entry of merged) {
-    console.log(`Speaker ${entry.speaker} (${entry.start.toFixed(2)}s - ${entry.end.toFixed(2)}s):`);
-    console.log(`  ${entry.text}\n`);
+catch (error) {
+    console.error("❌ Error:", error);
+    process.exit(1);
 }
-console.log("=".repeat(60));
-console.log("\nDone!");
 // ── Helpers ──
 function parseDiarization(text) {
     const segs = [];

package/dist/examples/transcription/parakeet-tdt-filesystem.js

@@ -1,28 +1,31 @@
-import { loadModel, unloadModel, transcribe, PARAKEET_TDT_ENCODER_FP32, PARAKEET_TDT_DECODER_FP32, PARAKEET_TDT_VOCAB, PARAKEET_TDT_PREPROCESSOR_FP32, } from "@qvac/sdk";
+/**
+ * Parakeet TDT transcription from a WAV file.
+ *
+ * Usage:
+ *   bun run examples/transcription/parakeet-tdt-filesystem.ts <wav-file> [parakeet-tdt-gguf]
+ *
+ * Loads a single GGUF checkpoint (`PARAKEET_TDT_0_6B_V3_Q8_0` by default) and
+ * transcribes the file with the batch `transcribe` API. Omit the model
+ * argument to use the registry constant.
+ *
+ * Audio should be 16 kHz mono PCM in a WAV container.
+ */
+import { loadModel, unloadModel, transcribe, PARAKEET_TDT_0_6B_V3_Q8_0, } from "@qvac/sdk";
 const args = process.argv.slice(2);
 if (!args[0]) {
     console.error("Usage: bun run examples/transcription/parakeet-tdt-filesystem.ts <wav-file-path> " +
-        "[encoder-onnx] [decoder-onnx] [vocab-txt] [preprocessor-onnx]");
-    console.error("\nIf model paths are omitted, defaults to registry models.");
+        "[parakeet-tdt-gguf]");
+    console.error("\nIf the model path is omitted, defaults to the registry model.");
     process.exit(1);
 }
 const audioFilePath = args[0];
-const parakeetEncoderSrc = args[1] ?? PARAKEET_TDT_ENCODER_FP32;
-const parakeetDecoderSrc = args[2] ?? PARAKEET_TDT_DECODER_FP32;
-const parakeetVocabSrc = args[3] ?? PARAKEET_TDT_VOCAB;
-const parakeetPreprocessorSrc = args[4] ?? PARAKEET_TDT_PREPROCESSOR_FP32;
+const parakeetModelSrc = args[1] ?? PARAKEET_TDT_0_6B_V3_Q8_0;
 try {
     console.log("Starting Parakeet transcription example...");
     console.log("Loading Parakeet model...");
     const modelId = await loadModel({
-        modelSrc: parakeetEncoderSrc,
+        modelSrc: parakeetModelSrc,
         modelType: "parakeet",
-        modelConfig: {
-            parakeetEncoderSrc,
-            parakeetDecoderSrc,
-            parakeetVocabSrc,
-            parakeetPreprocessorSrc,
-        },
         onProgress: (progress) => {
             console.log(`Download progress: ${progress.percentage.toFixed(1)}%`);
         },
@@ -37,6 +40,6 @@ try {
     console.log("Parakeet model unloaded successfully");
 }
 catch (error) {
-    console.error("Error:", error);
+    console.error("❌ Error:", error);
     process.exit(1);
 }

package/dist/examples/transcription/whispercpp-microphone-conversation.js

@@ -90,7 +90,12 @@ try {
                 }
                 break;
             case "endOfTurn":
-                console.log(`[endOfTurn] silence ${event.silenceDurationMs}ms — turn complete\n`);
+                if (event.source === "whisper") {
+                    console.log(`[endOfTurn] silence ${event.silenceDurationMs}ms — turn complete\n`);
+                }
+                else {
+                    console.log(`[endOfTurn] turn complete (token-driven EOU)\n`);
+                }
                 break;
         }
     }

package/dist/examples/tts/chatterbox.js

@@ -1,28 +1,19 @@
-import { loadModel, textToSpeech, unloadModel, TTS_TOKENIZER_EN_CHATTERBOX, TTS_SPEECH_ENCODER_EN_CHATTERBOX_FP32, TTS_EMBED_TOKENS_EN_CHATTERBOX_FP32, TTS_CONDITIONAL_DECODER_EN_CHATTERBOX_FP32, TTS_LANGUAGE_MODEL_EN_CHATTERBOX_FP32, } from "@qvac/sdk";
+import { loadModel, textToSpeech, unloadModel, TTS_T3_TURBO_EN_CHATTERBOX_Q8_0, TTS_S3GEN_EN_CHATTERBOX, } from "@qvac/sdk";
 import { createWav, playAudio, int16ArrayToBuffer, createWavHeader, } from "./utils.js";
-// Chatterbox TTS: voice cloning with reference audio.
-// Uses registry model constants - downloads automatically from QVAC Registry.
-// Only reference audio WAV needs to be provided by the user.
-// Usage: node chatterbox-filesystem.js <referenceAudioSrc>
+// Chatterbox TTS (GGML): voice cloning with optional reference audio.
+// Uses registry model constants — downloads automatically from QVAC Registry.
+// Usage: node chatterbox.ts [referenceAudioSrc]
 const [referenceAudioSrc] = process.argv.slice(2);
-if (!referenceAudioSrc) {
-    console.error("Usage: node chatterbox-filesystem.js <referenceAudioSrc>");
-    process.exit(1);
-}
 const CHATTERBOX_SAMPLE_RATE = 24000;
 try {
     const modelId = await loadModel({
-        modelSrc: TTS_TOKENIZER_EN_CHATTERBOX.src,
+        modelSrc: TTS_T3_TURBO_EN_CHATTERBOX_Q8_0.src,
         modelType: "tts",
         modelConfig: {
             ttsEngine: "chatterbox",
             language: "en",
-            ttsTokenizerSrc: TTS_TOKENIZER_EN_CHATTERBOX.src,
-            ttsSpeechEncoderSrc: TTS_SPEECH_ENCODER_EN_CHATTERBOX_FP32.src,
-            ttsEmbedTokensSrc: TTS_EMBED_TOKENS_EN_CHATTERBOX_FP32.src,
-            ttsConditionalDecoderSrc: TTS_CONDITIONAL_DECODER_EN_CHATTERBOX_FP32.src,
-            ttsLanguageModelSrc: TTS_LANGUAGE_MODEL_EN_CHATTERBOX_FP32.src,
-            referenceAudioSrc,
+            s3genModelSrc: TTS_S3GEN_EN_CHATTERBOX.src,
+            ...(referenceAudioSrc ? { referenceAudioSrc } : {}),
         },
         onProgress: (progress) => {
             console.log(progress);

package/dist/examples/tts/llm-to-tts-streaming.js

@@ -12,7 +12,7 @@
  * Usage:
  *   bun run examples/tts/llm-to-tts-streaming.ts
  */
-import { completion, loadModel, textToSpeechStream, unloadModel, LLAMA_3_2_1B_INST_Q4_0, TTS_SUPERTONIC2_OFFICIAL_TEXT_ENCODER_SUPERTONE_FP32, TTS_SUPERTONIC2_OFFICIAL_DURATION_PREDICTOR_SUPERTONE_FP32, TTS_SUPERTONIC2_OFFICIAL_VECTOR_ESTIMATOR_SUPERTONE_FP32, TTS_SUPERTONIC2_OFFICIAL_VOCODER_SUPERTONE_FP32, TTS_SUPERTONIC2_OFFICIAL_UNICODE_INDEXER_SUPERTONE_FP32, TTS_SUPERTONIC2_OFFICIAL_TTS_CONFIG_SUPERTONE, TTS_SUPERTONIC2_OFFICIAL_VOICE_STYLE_SUPERTONE, } from "@qvac/sdk";
+import { completion, loadModel, textToSpeechStream, unloadModel, LLAMA_3_2_1B_INST_Q4_0, TTS_EN_SUPERTONIC_Q8_0, } from "@qvac/sdk";
 import { createWav, playPcmInt16Chunk } from "./utils.js";
 const SUPERTONIC_SAMPLE_RATE = 44100;
 /**
@@ -69,21 +69,14 @@ try {
     console.log(`LLM ready: ${llmModelId}`);
     console.log("Loading Supertonic TTS (registry)…");
     const ttsModelId = await loadModel({
-        modelSrc: TTS_SUPERTONIC2_OFFICIAL_TEXT_ENCODER_SUPERTONE_FP32.src,
+        modelSrc: TTS_EN_SUPERTONIC_Q8_0.src,
         modelType: "tts",
         modelConfig: {
             ttsEngine: "supertonic",
             language: "en",
+            voice: "F1",
             ttsSpeed: 1.05,
             ttsNumInferenceSteps: 10,
-            ttsSupertonicMultilingual: true,
-            ttsTextEncoderSrc: TTS_SUPERTONIC2_OFFICIAL_TEXT_ENCODER_SUPERTONE_FP32.src,
-            ttsDurationPredictorSrc: TTS_SUPERTONIC2_OFFICIAL_DURATION_PREDICTOR_SUPERTONE_FP32.src,
-            ttsVectorEstimatorSrc: TTS_SUPERTONIC2_OFFICIAL_VECTOR_ESTIMATOR_SUPERTONE_FP32.src,
-            ttsVocoderSrc: TTS_SUPERTONIC2_OFFICIAL_VOCODER_SUPERTONE_FP32.src,
-            ttsUnicodeIndexerSrc: TTS_SUPERTONIC2_OFFICIAL_UNICODE_INDEXER_SUPERTONE_FP32.src,
-            ttsTtsConfigSrc: TTS_SUPERTONIC2_OFFICIAL_TTS_CONFIG_SUPERTONE.src,
-            ttsVoiceStyleSrc: TTS_SUPERTONIC2_OFFICIAL_VOICE_STYLE_SUPERTONE.src,
         },
         onProgress: (progress) => console.log(`TTS load: ${progress.percentage.toFixed(1)}%`),
     });

package/dist/examples/tts/supertonic-multilingual.js

@@ -1,25 +1,17 @@
-import { loadModel, textToSpeech, unloadModel, TTS_SUPERTONIC2_OFFICIAL_TEXT_ENCODER_SUPERTONE_FP32, TTS_SUPERTONIC2_OFFICIAL_DURATION_PREDICTOR_SUPERTONE_FP32, TTS_SUPERTONIC2_OFFICIAL_VECTOR_ESTIMATOR_SUPERTONE_FP32, TTS_SUPERTONIC2_OFFICIAL_VOCODER_SUPERTONE_FP32, TTS_SUPERTONIC2_OFFICIAL_TTS_CONFIG_SUPERTONE, TTS_SUPERTONIC2_OFFICIAL_VOICE_STYLE_SUPERTONE, TTS_SUPERTONIC2_OFFICIAL_UNICODE_INDEXER_SUPERTONE_FP32, } from "@qvac/sdk";
+import { loadModel, textToSpeech, unloadModel, TTS_MULTILINGUAL_SUPERTONIC2_Q8_0, } from "@qvac/sdk";
 import { createWav, playAudio, int16ArrayToBuffer, createWavHeader, } from "./utils.js";
-// Supertonic TTS: general-purpose, no voice cloning.
-// Uses registry model constants - downloads automatically from QVAC Registry.
+// Supertonic 2 TTS (GGML): multilingual synthesis (en/ko/es/pt/fr).
 const SUPERTONIC_SAMPLE_RATE = 44100;
 try {
     const modelId = await loadModel({
-        modelSrc: TTS_SUPERTONIC2_OFFICIAL_TEXT_ENCODER_SUPERTONE_FP32.src,
+        modelSrc: TTS_MULTILINGUAL_SUPERTONIC2_Q8_0.src,
         modelType: "tts",
         modelConfig: {
             ttsEngine: "supertonic",
             language: "es",
+            voice: "F1",
             ttsSpeed: 1.05,
             ttsNumInferenceSteps: 5,
-            ttsSupertonicMultilingual: true, //false for English quality.
-            ttsTextEncoderSrc: TTS_SUPERTONIC2_OFFICIAL_TEXT_ENCODER_SUPERTONE_FP32.src,
-            ttsDurationPredictorSrc: TTS_SUPERTONIC2_OFFICIAL_DURATION_PREDICTOR_SUPERTONE_FP32.src,
-            ttsVectorEstimatorSrc: TTS_SUPERTONIC2_OFFICIAL_VECTOR_ESTIMATOR_SUPERTONE_FP32.src,
-            ttsVocoderSrc: TTS_SUPERTONIC2_OFFICIAL_VOCODER_SUPERTONE_FP32.src,
-            ttsUnicodeIndexerSrc: TTS_SUPERTONIC2_OFFICIAL_UNICODE_INDEXER_SUPERTONE_FP32.src,
-            ttsTtsConfigSrc: TTS_SUPERTONIC2_OFFICIAL_TTS_CONFIG_SUPERTONE.src,
-            ttsVoiceStyleSrc: TTS_SUPERTONIC2_OFFICIAL_VOICE_STYLE_SUPERTONE.src,
         },
         onProgress: (progress) => {
             console.log(progress);

package/dist/examples/tts/supertonic.js

@@ -1,25 +1,18 @@
-import { loadModel, textToSpeech, unloadModel, TTS_SUPERTONIC2_OFFICIAL_TEXT_ENCODER_SUPERTONE_FP32, TTS_SUPERTONIC2_OFFICIAL_DURATION_PREDICTOR_SUPERTONE_FP32, TTS_SUPERTONIC2_OFFICIAL_VECTOR_ESTIMATOR_SUPERTONE_FP32, TTS_SUPERTONIC2_OFFICIAL_VOCODER_SUPERTONE_FP32, TTS_SUPERTONIC2_OFFICIAL_UNICODE_INDEXER_SUPERTONE_FP32, TTS_SUPERTONIC2_OFFICIAL_TTS_CONFIG_SUPERTONE, TTS_SUPERTONIC2_OFFICIAL_VOICE_STYLE_SUPERTONE, } from "@qvac/sdk";
+import { loadModel, textToSpeech, unloadModel, TTS_EN_SUPERTONIC_Q8_0, } from "@qvac/sdk";
 import { createWav, playAudio, int16ArrayToBuffer, createWavHeader, } from "./utils.js";
-// Supertonic TTS: general-purpose, no voice cloning.
-// Uses registry model constants - downloads automatically from QVAC Registry.
+// Supertonic TTS (GGML): fast English synthesis with baked-in voices.
+// Uses registry model constants — downloads automatically from QVAC Registry.
 const SUPERTONIC_SAMPLE_RATE = 44100;
 try {
     const modelId = await loadModel({
-        modelSrc: TTS_SUPERTONIC2_OFFICIAL_TEXT_ENCODER_SUPERTONE_FP32.src,
+        modelSrc: TTS_EN_SUPERTONIC_Q8_0.src,
         modelType: "tts",
         modelConfig: {
             ttsEngine: "supertonic",
             language: "en",
+            voice: "F1",
             ttsSpeed: 1.05,
             ttsNumInferenceSteps: 5,
-            ttsSupertonicMultilingual: true,
-            ttsTextEncoderSrc: TTS_SUPERTONIC2_OFFICIAL_TEXT_ENCODER_SUPERTONE_FP32.src,
-            ttsDurationPredictorSrc: TTS_SUPERTONIC2_OFFICIAL_DURATION_PREDICTOR_SUPERTONE_FP32.src,
-            ttsVectorEstimatorSrc: TTS_SUPERTONIC2_OFFICIAL_VECTOR_ESTIMATOR_SUPERTONE_FP32.src,
-            ttsVocoderSrc: TTS_SUPERTONIC2_OFFICIAL_VOCODER_SUPERTONE_FP32.src,
-            ttsUnicodeIndexerSrc: TTS_SUPERTONIC2_OFFICIAL_UNICODE_INDEXER_SUPERTONE_FP32.src,
-            ttsTtsConfigSrc: TTS_SUPERTONIC2_OFFICIAL_TTS_CONFIG_SUPERTONE.src,
-            ttsVoiceStyleSrc: TTS_SUPERTONIC2_OFFICIAL_VOICE_STYLE_SUPERTONE.src,
         },
         onProgress: (progress) => {
             console.log(progress);

package/dist/examples/vla-smolvla.d.ts

@@ -0,0 +1,2 @@
+export {};
+//# sourceMappingURL=vla-smolvla.d.ts.map

package/dist/examples/vla-smolvla.js

@@ -0,0 +1,75 @@
+/**
+ * SmolVLA (vision-language-action) example using the QVAC SDK.
+ *
+ * Loads the SmolVLA-LIBERO GGUF model, runs a single inference pass with
+ * synthetic inputs (zero-filled gray images + BOS-only tokens + zero state +
+ * zero noise), and prints the produced action chunk + per-stage timings.
+ *
+ * Usage:
+ *   bun examples/vla-smolvla.ts [path-to-smolvla.gguf]
+ *
+ * By default the example pulls the registry-baked SmolVLA-LIBERO GGUF
+ * (~1.9 GB) on first run and caches it locally. Pass an absolute path on
+ * the command line to override and load a local GGUF instead.
+ */
+import { close, loadModel, SMOLVLA_LIBERO_VISION_Q8, unloadModel, vla, vlaHparams, vlaPadState, vlaPreprocessImage, } from "@qvac/sdk";
+const modelSrcOverride = process.argv[2];
+const modelSrc = modelSrcOverride ?? SMOLVLA_LIBERO_VISION_Q8;
+try {
+    console.log("Loading SmolVLA model...");
+    const modelId = await loadModel({
+        modelSrc,
+        modelType: "vla",
+        modelConfig: { backend: "cpu" },
+        onProgress: (p) => typeof modelSrc === "string"
+            ? undefined
+            : process.stdout.write(`\rDownloading: ${p.percentage.toFixed(1)}%`),
+    });
+    if (typeof modelSrc !== "string")
+        process.stdout.write("\n");
+    console.log(`Model loaded: ${modelId}`);
+    const { hparams, backendName } = await vlaHparams({ modelId });
+    console.log(`Backend: ${backendName ?? "(unknown)"}`);
+    console.log("Hparams:", hparams);
+    // Build synthetic inputs sized to the model's expectations. A real
+    // consumer would: read camera frames, tokenize the instruction with the
+    // SmolVLM2 tokenizer, and read the robot's current end-effector pose.
+    const size = hparams.visionImageSize;
+    const dummyPixels = new Uint8Array(size * size * 3).fill(128);
+    const front = vlaPreprocessImage(dummyPixels, size, size, { size });
+    const wrist = vlaPreprocessImage(dummyPixels, size, size, { size });
+    const tokens = new Int32Array(hparams.tokenizerMaxLength);
+    const mask = new Uint8Array(hparams.tokenizerMaxLength);
+    // BOS-only "instruction" for the smoke test.
+    tokens[0] = 1;
+    mask[0] = 1;
+    const state = vlaPadState([0, 0, 0, 0, 0, 0], hparams.maxStateDim);
+    const noise = new Float32Array(hparams.chunkSize * hparams.maxActionDim);
+    console.log("Running VLA inference...");
+    const { actions, actionDim, chunkSize, stats } = await vla({
+        modelId,
+        images: [front, wrist],
+        imgWidth: size,
+        imgHeight: size,
+        state,
+        tokens,
+        mask,
+        noise,
+    });
+    console.log(`Got ${chunkSize} action steps of dim ${actionDim}.`);
+    console.log("First step:", Array.from(actions.subarray(0, actionDim)));
+    if (stats) {
+        console.log(`Timing: vision=${stats.vision_ms?.toFixed(0)}ms ` +
+            `smollm2=${stats.smollm2_total_ms?.toFixed(0)}ms ` +
+            `ode=${stats.ode_ms?.toFixed(0)}ms ` +
+            `total=${stats.total_ms?.toFixed(0)}ms`);
+    }
+    await unloadModel({ modelId, clearStorage: false });
+    console.log("Model unloaded.");
+    process.exit(0);
+}
+catch (error) {
+    console.error("VLA example failed:", error);
+    await close();
+    process.exit(1);
+}

package/dist/examples/voice-assistant/voice-assistant.js

@@ -10,7 +10,7 @@
  *
  * Requirements: FFmpeg installed, microphone access, speakers.
  */
-import { loadModel, unloadModel, transcribeStream, completion, textToSpeech, WHISPER_TINY, VAD_SILERO_5_1_2, LLAMA_3_2_1B_INST_Q4_0, TTS_SUPERTONIC2_OFFICIAL_TEXT_ENCODER_SUPERTONE_FP32, TTS_SUPERTONIC2_OFFICIAL_DURATION_PREDICTOR_SUPERTONE_FP32, TTS_SUPERTONIC2_OFFICIAL_VECTOR_ESTIMATOR_SUPERTONE_FP32, TTS_SUPERTONIC2_OFFICIAL_VOCODER_SUPERTONE_FP32, TTS_SUPERTONIC2_OFFICIAL_UNICODE_INDEXER_SUPERTONE_FP32, TTS_SUPERTONIC2_OFFICIAL_TTS_CONFIG_SUPERTONE, TTS_SUPERTONIC2_OFFICIAL_VOICE_STYLE_SUPERTONE, } from "@qvac/sdk";
+import { loadModel, unloadModel, transcribeStream, completion, textToSpeech, WHISPER_TINY, VAD_SILERO_5_1_2, LLAMA_3_2_1B_INST_Q4_0, TTS_EN_SUPERTONIC_Q8_0, } from "@qvac/sdk";
 import { spawnSync } from "child_process";
 import { startMicrophone } from "../audio/mic-input";
 import { createWavHeader, int16ArrayToBuffer, playAudio } from "../tts/utils";
@@ -100,21 +100,14 @@ const llmModelId = await loadModel({
 });
 console.log("Loading Supertonic TTS...");
 const ttsModelId = await loadModel({
-    modelSrc: TTS_SUPERTONIC2_OFFICIAL_TEXT_ENCODER_SUPERTONE_FP32.src,
+    modelSrc: TTS_EN_SUPERTONIC_Q8_0.src,
     modelType: "tts",
     modelConfig: {
         ttsEngine: "supertonic",
         language: "en",
+        voice: "F1",
         ttsSpeed: 1.05,
         ttsNumInferenceSteps: 5,
-        ttsSupertonicMultilingual: false,
-        ttsTextEncoderSrc: TTS_SUPERTONIC2_OFFICIAL_TEXT_ENCODER_SUPERTONE_FP32.src,
-        ttsDurationPredictorSrc: TTS_SUPERTONIC2_OFFICIAL_DURATION_PREDICTOR_SUPERTONE_FP32.src,
-        ttsVectorEstimatorSrc: TTS_SUPERTONIC2_OFFICIAL_VECTOR_ESTIMATOR_SUPERTONE_FP32.src,
-        ttsVocoderSrc: TTS_SUPERTONIC2_OFFICIAL_VOCODER_SUPERTONE_FP32.src,
-        ttsUnicodeIndexerSrc: TTS_SUPERTONIC2_OFFICIAL_UNICODE_INDEXER_SUPERTONE_FP32.src,
-        ttsTtsConfigSrc: TTS_SUPERTONIC2_OFFICIAL_TTS_CONFIG_SUPERTONE.src,
-        ttsVoiceStyleSrc: TTS_SUPERTONIC2_OFFICIAL_VOICE_STYLE_SUPERTONE.src,
     },
 });
 console.log("All models loaded.\n");

package/dist/expo/plugins/find-in-ancestor-node-modules.d.ts

@@ -0,0 +1,14 @@
+/**
+ * Walks from `startDir` up to the filesystem root, returning the first
+ * `<dir>/node_modules/<name>` that exists, or `null`. Matches Node's own
+ * module resolution so packages hoisted to a monorepo root (e.g. by bun or
+ * yarn workspaces) are found from a nested project directory.
+ *
+ * Use this as the building block for any plugin lookup that historically
+ * checked only `<projectRoot>/node_modules`. For SDK resolution specifically
+ * use `resolveSDKPackageDir`, which layers multi-package disambiguation and
+ * shadowed-copy warnings on top of this.
+ */
+declare function findInAncestorNodeModules(startDir: string, name: string): string | null;
+export { findInAncestorNodeModules };
+//# sourceMappingURL=find-in-ancestor-node-modules.d.ts.map

package/dist/expo/plugins/find-in-ancestor-node-modules.js

@@ -0,0 +1,24 @@
+import * as fs from "fs";
+import * as path from "path";
+/**
+ * Walks from `startDir` up to the filesystem root, returning the first
+ * `<dir>/node_modules/<name>` that exists, or `null`. Matches Node's own
+ * module resolution so packages hoisted to a monorepo root (e.g. by bun or
+ * yarn workspaces) are found from a nested project directory.
+ *
+ * Use this as the building block for any plugin lookup that historically
+ * checked only `<projectRoot>/node_modules`. For SDK resolution specifically
+ * use `resolveSDKPackageDir`, which layers multi-package disambiguation and
+ * shadowed-copy warnings on top of this.
+ */
+function findInAncestorNodeModules(startDir, name) {
+    let dir = startDir;
+    let parent = path.dirname(dir);
+    for (; dir !== parent; dir = parent, parent = path.dirname(dir)) {
+        const candidate = path.join(dir, "node_modules", name);
+        if (fs.existsSync(candidate))
+            return candidate;
+    }
+    return null;
+}
+export { findInAncestorNodeModules };

package/dist/expo/plugins/get-project-root.d.ts

@@ -0,0 +1,22 @@
+import type { ExportedConfigWithProps } from "@expo/config-plugins";
+import type { ExpoConfig } from "expo/config";
+/**
+ * `projectRoot` resolution in Expo plugins depends on plugin type:
+ *
+ * - **Mod plugins** (callbacks passed to `withDangerousMod` etc.) receive an
+ *   `ExportedConfigWithProps<...>` whose `modRequest.projectRoot` is the
+ *   documented public API for the consumer's project directory.
+ * - **Base plugins** (called with the bare `ExpoConfig`) run before any mod
+ *   cycle and have no `modRequest`. The only way to recover the project root
+ *   is the (undocumented but stable) `_internal.projectRoot` escape hatch
+ *   Expo's config-plugins runtime attaches to the config it hands plugins.
+ *
+ * Both helpers exist so the two access patterns are picked deliberately —
+ * call the one that matches the plugin shape you're in. Don't reach into the
+ * raw fields directly; that's how `withMobileBundle` and `withDeviceInfo`
+ * silently disagreed on which directory they were resolving against.
+ */
+declare function getProjectRootFromMod(config: ExportedConfigWithProps<unknown>): string;
+declare function getProjectRootFromBaseConfig(config: ExpoConfig): string | undefined;
+export { getProjectRootFromMod, getProjectRootFromBaseConfig };
+//# sourceMappingURL=get-project-root.d.ts.map

package/dist/expo/plugins/get-project-root.js

@@ -0,0 +1,23 @@
+/**
+ * `projectRoot` resolution in Expo plugins depends on plugin type:
+ *
+ * - **Mod plugins** (callbacks passed to `withDangerousMod` etc.) receive an
+ *   `ExportedConfigWithProps<...>` whose `modRequest.projectRoot` is the
+ *   documented public API for the consumer's project directory.
+ * - **Base plugins** (called with the bare `ExpoConfig`) run before any mod
+ *   cycle and have no `modRequest`. The only way to recover the project root
+ *   is the (undocumented but stable) `_internal.projectRoot` escape hatch
+ *   Expo's config-plugins runtime attaches to the config it hands plugins.
+ *
+ * Both helpers exist so the two access patterns are picked deliberately —
+ * call the one that matches the plugin shape you're in. Don't reach into the
+ * raw fields directly; that's how `withMobileBundle` and `withDeviceInfo`
+ * silently disagreed on which directory they were resolving against.
+ */
+function getProjectRootFromMod(config) {
+    return config.modRequest.projectRoot;
+}
+function getProjectRootFromBaseConfig(config) {
+    return config._internal?.["projectRoot"];
+}
+export { getProjectRootFromMod, getProjectRootFromBaseConfig };

package/dist/expo/plugins/index.d.ts

@@ -4,5 +4,9 @@ export { default as withDeviceInfo } from "./withDeviceInfo";
 export { default as withMobileBundle } from "./withMobileBundle";
 export { default as withOpenCL } from "./withOpenCL";
 export { default as withQvacSDK } from "./withQvacSDK";
+export { getProjectRootFromMod, getProjectRootFromBaseConfig, } from "./get-project-root";
+export { resolveSDKPackageDir, SDK_PACKAGE_NAMES, } from "./resolve-sdk-package-dir";
+export type { SDKPackageInfo } from "./resolve-sdk-package-dir";
+export { findInAncestorNodeModules } from "./find-in-ancestor-node-modules";
 export { default } from "./withQvacSDK";
 //# sourceMappingURL=index.d.ts.map

package/dist/expo/plugins/index.js

@@ -4,5 +4,9 @@ export { default as withDeviceInfo } from "./withDeviceInfo.js";
 export { default as withMobileBundle } from "./withMobileBundle.js";
 export { default as withOpenCL } from "./withOpenCL.js";
 export { default as withQvacSDK } from "./withQvacSDK.js";
+// Helpers for downstream plugin authors composing on top of these plugins.
+export { getProjectRootFromMod, getProjectRootFromBaseConfig, } from "./get-project-root.js";
+export { resolveSDKPackageDir, SDK_PACKAGE_NAMES, } from "./resolve-sdk-package-dir.js";
+export { findInAncestorNodeModules } from "./find-in-ancestor-node-modules.js";
 // Export the main plugin as default for easy usage
 export { default } from "./withQvacSDK.js";

package/dist/expo/plugins/resolve-sdk-package-dir.d.ts

@@ -6,8 +6,15 @@ type SDKPackageInfo = {
 /**
  * Resolves the installed SDK package directory from node_modules.
  *
- * Checks all known published package names and returns the one that exists.
- * Throws if none found or if multiple are found (ambiguous installation).
+ * Walks from `projectRoot` up to the filesystem root, checking each
+ * `node_modules` directory along the way. This matches Node's own module
+ * resolution so packages hoisted to a monorepo root (e.g. by bun or yarn
+ * workspaces) are found correctly.
+ *
+ * When multiple SDK installs are found across the tree, the one closest to
+ * `projectRoot` wins and a warning lists the others. Only throws
+ * `MultipleSDKInstallationsError` when two *different* SDK package names
+ * coexist at the same nesting level (a real ambiguity in package selection).
  */
 declare function resolveSDKPackageDir(projectRoot: string): SDKPackageInfo;
 export { resolveSDKPackageDir, SDK_PACKAGE_NAMES };