noosphere 0.9.1 → 0.9.2

package/dist/index.cjs

@@ -2716,6 +2716,36 @@ var OpenAIMediaProvider = class {
   modalities = ["image", "video", "tts", "stt"];
   isLocal = false;
   modelsCache = null;
+  voicesCache = null;
+  /** Auto-fetch available TTS voices by sending an invalid voice and parsing the error. */
+  async fetchVoices() {
+    if (this.voicesCache) return this.voicesCache;
+    try {
+      const res = await fetch(`${OPENAI_API_BASE}/audio/speech`, {
+        method: "POST",
+        headers: {
+          "Content-Type": "application/json",
+          Authorization: `Bearer ${this.apiKey}`
+        },
+        body: JSON.stringify({ model: "tts-1", input: ".", voice: "__discover_voices__" })
+      });
+      if (!res.ok) {
+        const data = await res.json();
+        const msg = data?.error?.message ?? "";
+        const shouldBe = msg.match(/Input should be ([^"]+)/);
+        if (shouldBe) {
+          const voiceList = shouldBe[1].match(/'([a-z]+)'/g);
+          if (voiceList && voiceList.length > 0) {
+            this.voicesCache = voiceList.map((v) => v.replace(/'/g, ""));
+            return this.voicesCache;
+          }
+        }
+      }
+    } catch {
+    }
+    this.voicesCache = [];
+    return this.voicesCache;
+  }
   async ping() {
     try {
       const controller = new AbortController();
@@ -2751,6 +2781,7 @@ var OpenAIMediaProvider = class {
       } finally {
         clearTimeout(timer);
       }
+      const voices = await this.fetchVoices();
       const entries = data?.data ?? [];
       const logo = getProviderLogo("openai");
       const models = [];
@@ -2766,7 +2797,7 @@ var OpenAIMediaProvider = class {
           cost: { price: 0, unit: "per_request" },
           logo,
           description: entry.description,
-          capabilities: this.getCapabilities(entry.id, mod)
+          capabilities: this.getCapabilities(entry.id, mod, voices)
         };
         models.push(info);
       }
@@ -2911,7 +2942,7 @@ var OpenAIMediaProvider = class {
       }
     };
   }
-  getCapabilities(id, modality) {
+  getCapabilities(id, modality, voices) {
     if (modality === "image") {
       return {
         maxWidth: id.startsWith("dall-e-3") ? 1792 : 1024,
@@ -2920,7 +2951,7 @@ var OpenAIMediaProvider = class {
     }
     if (modality === "tts") {
       return {
-        voices: ["alloy", "ash", "coral", "echo", "fable", "onyx", "nova", "sage", "shimmer"]
+        voices: voices && voices.length > 0 ? voices : void 0
       };
     }
     if (modality === "video") {
@@ -2941,18 +2972,34 @@ var OpenAIMediaProvider = class {
 // src/providers/google-media.ts
 var GOOGLE_API_BASE = "https://generativelanguage.googleapis.com/v1beta";
 var FETCH_TIMEOUT_MS6 = 8e3;
-var GOOGLE_TTS_VOICES = [
-  "Aoede",
-  "Charon",
-  "Fenrir",
-  "Kore",
-  "Puck",
-  "Leda",
-  "Orus",
-  "Perseus",
-  "Zephyr",
-  "Callirrhoe"
-];
+async function fetchGoogleVoices(apiKey) {
+  try {
+    const res = await fetch(
+      `${GOOGLE_API_BASE}/models/gemini-2.5-flash-preview-tts:generateContent?key=${apiKey}`,
+      {
+        method: "POST",
+        headers: { "Content-Type": "application/json" },
+        body: JSON.stringify({
+          contents: [{ parts: [{ text: "." }] }],
+          generationConfig: {
+            response_modalities: ["AUDIO"],
+            speech_config: { voiceConfig: { prebuiltVoiceConfig: { voiceName: "__discover_voices__" } } }
+          }
+        })
+      }
+    );
+    if (!res.ok) {
+      const data = await res.json();
+      const msg = data?.error?.message ?? "";
+      const match = msg.match(/Allowed voice names are:\s*(.+)/i);
+      if (match) {
+        return match[1].split(",").map((v) => v.trim()).filter(Boolean);
+      }
+    }
+  } catch {
+  }
+  return [];
+}
 function classifyGoogleModel(model) {
   const name = (model.name ?? "").replace("models/", "");
   const methods = model.supportedGenerationMethods ?? [];
@@ -2970,6 +3017,7 @@ var GoogleMediaProvider = class {
   modalities = ["image", "video", "tts"];
   isLocal = false;
   modelsCache = null;
+  voicesCache = null;
   async ping() {
     try {
       const controller = new AbortController();
@@ -3004,6 +3052,9 @@ var GoogleMediaProvider = class {
         clearTimeout(timer);
       }
       const entries = data?.models ?? [];
+      if (!this.voicesCache) {
+        this.voicesCache = await fetchGoogleVoices(this.apiKey);
+      }
       const logo = getProviderLogo("google");
       const models = [];
       for (const entry of entries) {
@@ -3020,7 +3071,7 @@ var GoogleMediaProvider = class {
           cost: { price: 0, unit: modality2 === "video" ? "per_video" : "per_image" },
           logo,
           description: entry.description,
-          capabilities: modality2 === "video" ? { maxDuration: 8, supportsStreaming: false } : modality2 === "tts" ? { voices: GOOGLE_TTS_VOICES } : void 0
+          capabilities: modality2 === "video" ? { maxDuration: 8, supportsStreaming: false } : modality2 === "tts" ? { voices: this.voicesCache && this.voicesCache.length > 0 ? this.voicesCache : void 0 } : void 0
         };
         models.push(info);
       }