makefx 1.6.1 → 1.6.3

package/README.md

@@ -61,4 +61,4 @@ IDs, or runs. Retired commands and aliases are not redirected.
 
 Run `makefx --help` or `makefx <noun> --help` for current options.
 
-Version: 1.6.1
+Version: 1.6.3

package/makefx.mjs

@@ -4438,7 +4438,7 @@ require_websocket_server();
 var wrapper_default = import_websocket.default;
 //#endregion
 //#region src/cli/version.ts
-var CLI_VERSION = "1.6.1+00259d259e65";
+var CLI_VERSION = "1.6.3+203ed3ea216e";
 var CLI_VERSION_HEADER = "X-MakeFX-CLI-Version";
 function cliVersionHeaders() {
 	return {
@@ -5780,6 +5780,18 @@ Usage:
 `);
 }
 //#endregion
+//#region src/shared/audioGenerationOptions.ts
+var DEFAULT_ELEVENLABS_SPEECH_MODEL_ID = "eleven_v3";
+/**
+* Keep the provider-native ElevenLabs ID canonical everywhere after an input
+* boundary. The hyphenated MakeFX slug is accepted only for migration.
+*/
+function normalizeElevenLabsSpeechModelId(model) {
+	const normalized = model?.trim();
+	if (!normalized) return void 0;
+	return normalized === "eleven-v3" ? DEFAULT_ELEVENLABS_SPEECH_MODEL_ID : normalized;
+}
+//#endregion
 //#region src/backend/utils/audio-metadata.ts
 function readWavDurationMs(bytes) {
 	if (bytes.byteLength < 12) return void 0;
@@ -6870,8 +6882,9 @@ function getAudioModeFromType(type) {
 }
 function validateAudioModelOption(value, mode) {
 	if (!value) return;
-	const supported = mode === "music" ? ["eleven-music", "lyria-3"] : mode === "sfx" ? ["eleven-sfx"] : mode === "speech" || mode === "dialogue" ? ["eleven-v3"] : [];
-	if (!supported.includes(value)) throw new Error(`Unsupported ${mode ?? "audio"} model: ${value}. Expected ${supported.join(" or ")}`);
+	const normalized = mode === "speech" || mode === "dialogue" ? normalizeElevenLabsSpeechModelId(value) : value;
+	const supported = mode === "music" ? ["eleven-music", "lyria-3"] : mode === "sfx" ? ["eleven-sfx"] : mode === "speech" || mode === "dialogue" ? [DEFAULT_ELEVENLABS_SPEECH_MODEL_ID] : [];
+	if (!normalized || !supported.includes(normalized)) throw new Error(`Unsupported ${mode ?? "audio"} model: ${value}. Expected ${supported.join(" or ")}`);
 }
 function parseMusicProviderOption(parsed, mediaKind, assetType) {
 	const value = parsed.options.model;
@@ -7067,7 +7080,6 @@ async function preflightLocalSeedanceReferences(options, deps, state) {
 	if (videoBytes >= 50 * 1024 * 1024) throw new Error("Seedance reference videos must total under 50 MB");
 	const hasCompleteVideoDurations = inspections.video.every((item) => item.durationMs !== void 0);
 	if (inspections.video.length > 0 && hasCompleteVideoDurations && (videoDurationMs < 2e3 || videoDurationMs > 15e3)) throw new Error("Seedance reference videos must total 2 to 15 seconds");
-	if (inspections.audio.reduce((total, item) => total + (item.durationMs ?? 0), 0) > 15e3) throw new Error("Seedance reference audio must total at most 15 seconds");
 }
 async function resolveReferenceVariantIds(refs, ctx, deps, variants = [], mediaKind = CLI_GENERATION_MEDIA_KIND, assets = [], expectedReferenceMediaKind) {
 	const variantIds = [];
@@ -7435,9 +7447,12 @@ function normalizeCliOption(value) {
 	if (value === void 0 || value === "true") return void 0;
 	return value.trim() || void 0;
 }
-function parseAudioGenerationOptions(_parsed, mediaKind) {
+function parseAudioGenerationOptions(parsed, mediaKind) {
 	if (mediaKind !== "audio") return {};
-	return {};
+	const mode = getAudioModeFromType(parsed.options.type);
+	if (mode !== "speech" && mode !== "dialogue") return {};
+	const model = normalizeElevenLabsSpeechModelId(parsed.options.model);
+	return model ? { model } : {};
 }
 function parseImageGenerationOptions(parsed, mediaKind) {
 	if (mediaKind !== "image") return {};
@@ -7638,6 +7653,7 @@ function parseWaitSeconds$1(value) {
 function buildRegenerationBody(mediaKind, parsed, referenceVariantIds, audioMode, sourceRecipe) {
 	const prompt = parsed.positionals[1] ?? parsed.options.prompt;
 	const model = parsed.options.model;
+	const audioModel = mediaKind === "audio" && (audioMode === "speech" || audioMode === "dialogue") ? normalizeElevenLabsSpeechModelId(model) : void 0;
 	const params = {};
 	let duration;
 	let modelProvider;
@@ -7658,7 +7674,7 @@ function buildRegenerationBody(mediaKind, parsed, referenceVariantIds, audioMode
 		if (parsed.options["no-audio"] === "true") params.generateAudio = false;
 	} else {
 		if (!audioMode) throw new Error("Audio regeneration mode is required");
-		validateAudioModel(audioMode, model);
+		validateAudioModel(audioMode, audioModel ?? model);
 		if (parsed.options.voice) params.voiceId = parsed.options.voice;
 		if (parsed.options["dialogue-voices"]) params.dialogueVoiceIds = splitList(parsed.options["dialogue-voices"]);
 		if (audioMode === "music" && model) params.musicProvider = model === "lyria-3" ? "lyria" : "elevenlabs";
@@ -7667,6 +7683,7 @@ function buildRegenerationBody(mediaKind, parsed, referenceVariantIds, audioMode
 		...prompt ? { prompt } : {},
 		...model && mediaKind === "image" ? { model } : {},
 		...model && mediaKind === "video" ? { model: resolveVideoRegenerationModel(model, parsed, sourceRecipe) } : {},
+		...audioModel ? { model: audioModel } : {},
 		...modelProvider ? { modelProvider } : {},
 		...referenceVariantIds !== void 0 ? { referenceVariantIds } : {},
 		...Object.keys(params).length > 0 ? { params } : {},
@@ -7708,7 +7725,7 @@ function parseStoredRecipe$2(recipe) {
 }
 function validateAudioModel(mode, model) {
 	if (!model) return;
-	const supported = mode === "music" ? ["eleven-music", "lyria-3"] : mode === "sfx" ? ["eleven-sfx"] : ["eleven-v3"];
+	const supported = mode === "music" ? ["eleven-music", "lyria-3"] : mode === "sfx" ? ["eleven-sfx"] : [DEFAULT_ELEVENLABS_SPEECH_MODEL_ID];
 	if (!supported.includes(model)) throw new Error(`Unsupported ${mode} model: ${model}. Expected ${supported.join(" or ")}`);
 }
 function parseReferences(mediaKind, options) {
@@ -11497,10 +11514,10 @@ Seedance mode is inferred: first/last frame selects frame mode, image/video/audi
   makefx audio voices [--json]
   makefx audio align <variant-ref> "transcript" [--space <id>] [--json]
   makefx audio align <variant-ref> --input transcript.txt [--space <id>] [--json]
-  makefx audio speech generate "text" --voice <voice-id> --name <name> -o <file>
-  makefx audio speech regenerate <variant-ref> ["text"] [--voice <voice-id>] [--no-activate] [--wait]
-  makefx audio dialogue generate --input <file> --dialogue-voices <id,id> --name <name> -o <file>
-  makefx audio dialogue regenerate <variant-ref> ["script"] [--dialogue-voices <id,id>] [--no-activate] [--wait]
+  makefx audio speech generate "text" [--model eleven_v3] --voice <voice-id> --name <name> -o <file>
+  makefx audio speech regenerate <variant-ref> ["text"] [--model eleven_v3] [--voice <voice-id>] [--no-activate] [--wait]
+  makefx audio dialogue generate --input <file> [--model eleven_v3] --dialogue-voices <id,id> --name <name> -o <file>
+  makefx audio dialogue regenerate <variant-ref> ["script"] [--model eleven_v3] [--dialogue-voices <id,id>] [--no-activate] [--wait]
   makefx audio music generate "prompt" --model eleven-music|lyria-3 --name <name> -o <file>
   makefx audio music regenerate <variant-ref> ["prompt"] [--model eleven-music|lyria-3] [--no-activate] [--wait]
   makefx audio sfx generate "prompt" --name <name> -o <file>

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "makefx",
-  "version": "1.6.1",
+  "version": "1.6.3",
   "description": "Command-line interface for AI-assisted game asset production with Make Effects.",
   "license": "MIT",
   "type": "module",