@amir83nasr/10router 0.1.0

package/open-sse/handlers/ttsProviders/gemini.js

@@ -0,0 +1,136 @@
+// Gemini TTS — generateContent with AUDIO modality returns PCM L16, wrap as WAV
+import { Buffer } from "node:buffer";
+import { PROVIDER_MEDIA, PROVIDER_MODELS } from "../../providers/index.js";
+
+const TTS_CFG = PROVIDER_MEDIA["gemini"]?.ttsConfig || {};
+const TTS_BASE = TTS_CFG.baseUrl;
+const FALLBACK_MODEL = "gemini-3.1-flash-tts-preview";
+const KNOWN_MODELS = [
+  ...(TTS_CFG.models || []),
+  ...(PROVIDER_MODELS["gemini-tts-models"] || []),
+  ...(PROVIDER_MODELS.gemini || []).filter((m) => (m.kind || m.type) === "tts"),
+]
+  .map((m) => m?.id)
+  .filter(Boolean)
+  .filter((id, index, list) => list.indexOf(id) === index);
+const DEFAULT_MODEL = KNOWN_MODELS[0] || FALLBACK_MODEL;
+const DEFAULT_VOICE = "Kore";
+
+// Parse "model/voice" — if input doesn't match a known TTS model, treat it as voice with default model
+function parseGeminiModelVoice(input) {
+  if (!input) return { modelId: DEFAULT_MODEL, voiceId: DEFAULT_VOICE };
+  for (const id of KNOWN_MODELS) {
+    if (input === id) return { modelId: id, voiceId: DEFAULT_VOICE };
+    if (input.startsWith(`${id}/`)) return { modelId: id, voiceId: input.slice(id.length + 1) };
+  }
+  return { modelId: DEFAULT_MODEL, voiceId: input };
+}
+// Gemini returns PCM 16-bit signed mono @ 24kHz
+const SAMPLE_RATE = 24000;
+const CHANNELS = 1;
+const BITS_PER_SAMPLE = 16;
+
+// Build WAV header for raw PCM payload
+function pcmToWav(pcmBuffer) {
+  const dataSize = pcmBuffer.length;
+  const byteRate = (SAMPLE_RATE * CHANNELS * BITS_PER_SAMPLE) / 8;
+  const blockAlign = (CHANNELS * BITS_PER_SAMPLE) / 8;
+  const header = Buffer.alloc(44);
+  header.write("RIFF", 0);
+  header.writeUInt32LE(36 + dataSize, 4);
+  header.write("WAVE", 8);
+  header.write("fmt ", 12);
+  header.writeUInt32LE(16, 16);
+  header.writeUInt16LE(1, 20);
+  header.writeUInt16LE(CHANNELS, 22);
+  header.writeUInt32LE(SAMPLE_RATE, 24);
+  header.writeUInt32LE(byteRate, 28);
+  header.writeUInt16LE(blockAlign, 32);
+  header.writeUInt16LE(BITS_PER_SAMPLE, 34);
+  header.write("data", 36);
+  header.writeUInt32LE(dataSize, 40);
+  return Buffer.concat([header, pcmBuffer]);
+}
+
+// Build TTS prompt: add "Say [in {language}]:" prefix to force TTS mode
+function buildPrompt(text, language) {
+  if (/:\s/.test(text)) return text; // user already provided style instruction
+  return language ? `Say in ${language}: ${text}` : `Say: ${text}`;
+}
+
+export default {
+  async synthesize(text, model, credentials, _responseFormat, opts = {}) {
+    if (!credentials?.apiKey) throw new Error("No Gemini API key configured");
+    const { modelId, voiceId } = parseGeminiModelVoice(model);
+    const url = `${TTS_BASE}/${modelId}:generateContent?key=${credentials.apiKey}`;
+    const res = await fetch(url, {
+      method: "POST",
+      headers: { "Content-Type": "application/json" },
+      body: JSON.stringify({
+        contents: [{ parts: [{ text: buildPrompt(text, opts.language) }] }],
+        generationConfig: {
+          responseModalities: ["AUDIO"],
+          speechConfig: { voiceConfig: { prebuiltVoiceConfig: { voiceName: voiceId } } },
+        },
+      }),
+    });
+    if (!res.ok) {
+      const err = await res.json().catch(() => ({}));
+      throw new Error(err?.error?.message || `Gemini TTS failed: ${res.status}`);
+    }
+    const data = await res.json();
+    const b64 = data?.candidates?.[0]?.content?.parts?.find((p) => p.inlineData?.data)?.inlineData
+      ?.data;
+    if (!b64) {
+      const reason =
+        data?.candidates?.[0]?.finishReason || data?.promptFeedback?.blockReason || "unknown";
+      throw new Error(
+        `Gemini TTS returned no audio (finishReason: ${reason}, voice: ${voiceId}, model: ${modelId})`,
+      );
+    }
+    const wav = pcmToWav(Buffer.from(b64, "base64"));
+    return { base64: wav.toString("base64"), format: "wav" };
+  },
+};
+
+// Voice fetcher — return prebuilt voices (Gemini has no list API)
+const PREBUILT_VOICES = [
+  { id: "Zephyr", lang: "en", gender: "Female" },
+  { id: "Puck", lang: "en", gender: "Male" },
+  { id: "Charon", lang: "en", gender: "Male" },
+  { id: "Kore", lang: "en", gender: "Female" },
+  { id: "Fenrir", lang: "en", gender: "Male" },
+  { id: "Leda", lang: "en", gender: "Female" },
+  { id: "Orus", lang: "en", gender: "Male" },
+  { id: "Aoede", lang: "en", gender: "Female" },
+  { id: "Callirrhoe", lang: "en", gender: "Female" },
+  { id: "Autonoe", lang: "en", gender: "Female" },
+  { id: "Enceladus", lang: "en", gender: "Male" },
+  { id: "Iapetus", lang: "en", gender: "Male" },
+  { id: "Umbriel", lang: "en", gender: "Male" },
+  { id: "Algieba", lang: "en", gender: "Male" },
+  { id: "Despina", lang: "en", gender: "Female" },
+  { id: "Erinome", lang: "en", gender: "Female" },
+  { id: "Algenib", lang: "en", gender: "Male" },
+  { id: "Rasalgethi", lang: "en", gender: "Male" },
+  { id: "Laomedeia", lang: "en", gender: "Female" },
+  { id: "Achernar", lang: "en", gender: "Female" },
+  { id: "Alnilam", lang: "en", gender: "Male" },
+  { id: "Schedar", lang: "en", gender: "Male" },
+  { id: "Gacrux", lang: "en", gender: "Female" },
+  { id: "Pulcherrima", lang: "en", gender: "Female" },
+  { id: "Achird", lang: "en", gender: "Male" },
+  { id: "Zubenelgenubi", lang: "en", gender: "Male" },
+  { id: "Vindemiatrix", lang: "en", gender: "Female" },
+  { id: "Sadachbia", lang: "en", gender: "Male" },
+  { id: "Sadaltager", lang: "en", gender: "Male" },
+  { id: "Sulafat", lang: "en", gender: "Female" },
+];
+
+export async function fetchGeminiVoices() {
+  return PREBUILT_VOICES.map((v) => ({
+    voice_id: v.id,
+    name: v.id,
+    labels: { language: v.lang, gender: v.gender },
+  }));
+}

package/open-sse/handlers/ttsProviders/genericFormats.js

@@ -0,0 +1,191 @@
+// Generic config-driven TTS handlers — dispatched by ttsConfig.format.
+// Each handler accepts { baseUrl, apiKey, text, modelId, voiceId } and returns { base64, format }.
+import { responseToBase64, throwUpstreamError } from "./_base.js";
+import minimaxTts from "./minimax.js";
+
+// Hyperbolic: POST { text } → { audio: base64 }
+async function hyperbolic({ baseUrl, apiKey, text }) {
+  const res = await fetch(baseUrl, {
+    method: "POST",
+    headers: { "Content-Type": "application/json", Authorization: `Bearer ${apiKey}` },
+    body: JSON.stringify({ text }),
+  });
+  if (!res.ok) await throwUpstreamError(res);
+  const data = await res.json();
+  return { base64: data.audio, format: "mp3" };
+}
+
+// Deepgram: model via query, Token auth, returns binary
+async function deepgram({ baseUrl, apiKey, text, modelId }) {
+  const url = new URL(baseUrl);
+  url.searchParams.set("model", modelId || "aura-asteria-en");
+  const res = await fetch(url.toString(), {
+    method: "POST",
+    headers: { "Content-Type": "application/json", Authorization: `Token ${apiKey}` },
+    body: JSON.stringify({ text }),
+  });
+  if (!res.ok) await throwUpstreamError(res);
+  return responseToBase64(res, "mp3");
+}
+
+// Nvidia NIM: POST { input: { text }, voice, model } → binary
+async function nvidia({ baseUrl, apiKey, text, modelId, voiceId }) {
+  const res = await fetch(baseUrl, {
+    method: "POST",
+    headers: { "Content-Type": "application/json", Authorization: `Bearer ${apiKey}` },
+    body: JSON.stringify({ input: { text }, voice: voiceId || "default", model: modelId }),
+  });
+  if (!res.ok) await throwUpstreamError(res);
+  return responseToBase64(res, "wav");
+}
+
+// HuggingFace: POST {baseUrl}/{modelId} { inputs: text } → binary
+async function huggingface({ baseUrl, apiKey, text, modelId }) {
+  if (!modelId || modelId.includes("..")) throw new Error("Invalid HuggingFace model ID");
+  const res = await fetch(`${baseUrl}/${modelId}`, {
+    method: "POST",
+    headers: { "Content-Type": "application/json", Authorization: `Bearer ${apiKey}` },
+    body: JSON.stringify({ inputs: text }),
+  });
+  if (!res.ok) await throwUpstreamError(res);
+  return responseToBase64(res, "wav");
+}
+
+// Fish Audio: model travels in an HTTP header, the voice is a reference_id, returns binary
+async function fishAudio({ baseUrl, apiKey, text, modelId, voiceId }) {
+  const res = await fetch(baseUrl, {
+    method: "POST",
+    headers: {
+      "Content-Type": "application/json",
+      Authorization: `Bearer ${apiKey}`,
+      model: modelId || "s2.1-pro-free",
+    },
+    body: JSON.stringify({
+      text,
+      format: "mp3",
+      ...(voiceId ? { reference_id: voiceId } : {}),
+    }),
+  });
+  if (!res.ok) await throwUpstreamError(res);
+  return responseToBase64(res, "mp3");
+}
+
+// Inworld: Basic auth, JSON { audioContent }
+async function inworld({ baseUrl, apiKey, text, modelId, voiceId }) {
+  const res = await fetch(baseUrl, {
+    method: "POST",
+    headers: { "Content-Type": "application/json", Authorization: `Basic ${apiKey}` },
+    body: JSON.stringify({
+      text,
+      voiceId: voiceId || "Alex",
+      modelId: modelId || "inworld-tts-1.5-mini",
+      audioConfig: { audioEncoding: "MP3" },
+    }),
+  });
+  if (!res.ok) await throwUpstreamError(res);
+  const data = await res.json();
+  if (!data.audioContent) throw new Error("Inworld TTS returned no audio");
+  return { base64: data.audioContent, format: "mp3" };
+}
+
+// Cartesia: X-API-Key header
+async function cartesia({ baseUrl, apiKey, text, modelId, voiceId }) {
+  const res = await fetch(baseUrl, {
+    method: "POST",
+    headers: {
+      "Content-Type": "application/json",
+      "X-API-Key": apiKey,
+      "Cartesia-Version": "2024-06-10",
+    },
+    body: JSON.stringify({
+      model_id: modelId || "sonic-2",
+      transcript: text,
+      ...(voiceId ? { voice: { mode: "id", id: voiceId } } : {}),
+      output_format: { container: "mp3", bit_rate: 128000, sample_rate: 44100 },
+    }),
+  });
+  if (!res.ok) await throwUpstreamError(res);
+  return responseToBase64(res, "mp3");
+}
+
+// PlayHT: token format "userId:apiKey", voice = s3 URL
+async function playht({ baseUrl, apiKey, text, modelId, voiceId }) {
+  const [userId, key] = (apiKey || ":").split(":");
+  const res = await fetch(baseUrl, {
+    method: "POST",
+    headers: {
+      "Content-Type": "application/json",
+      Accept: "audio/mpeg",
+      "X-USER-ID": userId || "",
+      Authorization: `Bearer ${key || apiKey}`,
+    },
+    body: JSON.stringify({
+      text,
+      voice:
+        voiceId ||
+        "s3://voice-cloning-zero-shot/d9ff78ba-d016-47f6-b0ef-dd630f59414e/female-cs/manifest.json",
+      voice_engine: modelId || "PlayDialog",
+      output_format: "mp3",
+      speed: 1,
+    }),
+  });
+  if (!res.ok) await throwUpstreamError(res);
+  return responseToBase64(res, "mp3");
+}
+
+// Coqui (local, noAuth): POST { text, speaker_id } → WAV
+async function coqui({ baseUrl, text, voiceId }) {
+  const res = await fetch(baseUrl, {
+    method: "POST",
+    headers: { "Content-Type": "application/json" },
+    body: JSON.stringify({ text, ...(voiceId ? { speaker_id: voiceId } : {}) }),
+  });
+  if (!res.ok) await throwUpstreamError(res);
+  return responseToBase64(res, "wav");
+}
+
+// Tortoise (local, noAuth)
+async function tortoise({ baseUrl, text, voiceId }) {
+  const res = await fetch(baseUrl, {
+    method: "POST",
+    headers: { "Content-Type": "application/json" },
+    body: JSON.stringify({ text, voice: voiceId || "random" }),
+  });
+  if (!res.ok) await throwUpstreamError(res);
+  return responseToBase64(res, "wav");
+}
+
+// OpenAI-compatible upstream (qwen3-tts, etc.)
+async function openaiCompat({ baseUrl, apiKey, text, modelId, voiceId }) {
+  const headers = { "Content-Type": "application/json" };
+  if (apiKey) headers["Authorization"] = `Bearer ${apiKey}`;
+  const res = await fetch(baseUrl, {
+    method: "POST",
+    headers,
+    body: JSON.stringify({
+      model: modelId,
+      input: text,
+      voice: voiceId || "alloy",
+      response_format: "mp3",
+      speed: 1.0,
+    }),
+  });
+  if (!res.ok) await throwUpstreamError(res);
+  return responseToBase64(res, "mp3");
+}
+
+// format → handler dispatcher
+export const FORMAT_HANDLERS = {
+  hyperbolic,
+  deepgram,
+  "nvidia-tts": nvidia,
+  "huggingface-tts": huggingface,
+  inworld,
+  cartesia,
+  playht,
+  coqui,
+  tortoise,
+  openai: openaiCompat,
+  "minimax-tts": minimaxTts,
+  "fish-audio": fishAudio,
+};

package/open-sse/handlers/ttsProviders/googleTts.js

@@ -0,0 +1,64 @@
+// Google Translate TTS (no auth) — scrape token + batchexecute RPC
+import { UA } from "./_base.js";
+
+const REFRESH_MS = 11 * 60 * 1000;
+const cache = { token: null, tokenTime: 0 };
+let _idx = 0;
+
+async function getToken() {
+  const now = Date.now();
+  if (cache.token && now - cache.tokenTime < REFRESH_MS) return cache.token;
+  const res = await fetch("https://translate.google.com/", { headers: { "User-Agent": UA } });
+  if (!res.ok) throw new Error(`Google translate fetch failed: ${res.status}`);
+  const html = await res.text();
+  const fSid = html.match(/"FdrFJe":"(.*?)"/)?.[1];
+  const bl = html.match(/"cfb2h":"(.*?)"/)?.[1];
+  if (!fSid || !bl) throw new Error("Failed to parse Google token");
+  cache.token = { "f.sid": fSid, bl };
+  cache.tokenTime = now;
+  return cache.token;
+}
+
+export default {
+  noAuth: true,
+  async synthesize(text, model) {
+    const lang = model || "en";
+    const token = await getToken();
+    const cleanText = text
+      .replace(/[@^*()\\/\-_+=><"'\u201c\u201d\u3010\u3011]/g, " ")
+      .replaceAll(", ", ". ");
+    const rpcId = "jQ1olc";
+    const reqId = ++_idx * 100000 + Math.floor(1000 + Math.random() * 9000);
+    const query = new URLSearchParams({
+      rpcids: rpcId,
+      "f.sid": token["f.sid"],
+      bl: token.bl,
+      hl: lang,
+      "soc-app": 1,
+      "soc-platform": 1,
+      "soc-device": 1,
+      _reqid: reqId,
+      rt: "c",
+    });
+    const payload = [cleanText, lang, null, "undefined", [0]];
+    const body = new URLSearchParams();
+    body.append("f.req", JSON.stringify([[[rpcId, JSON.stringify(payload), null, "generic"]]]));
+    const res = await fetch(
+      `https://translate.google.com/_/TranslateWebserverUi/data/batchexecute?${query}`,
+      {
+        method: "POST",
+        headers: {
+          "Content-Type": "application/x-www-form-urlencoded",
+          Referer: "https://translate.google.com/",
+        },
+        body: body.toString(),
+      },
+    );
+    if (!res.ok) throw new Error(`Google TTS failed: ${res.status}`);
+    const data = await res.text();
+    const split = JSON.parse(data.split("\n")[3]);
+    const base64 = JSON.parse(split[0][2])[0];
+    if (!base64 || base64.length < 100) throw new Error("Google TTS returned empty audio");
+    return { base64, format: "mp3" };
+  },
+};

package/open-sse/handlers/ttsProviders/index.js

@@ -0,0 +1,56 @@
+// TTS provider registry
+import googleTts from "./googleTts.js";
+import edgeTts, { fetchEdgeTtsVoices } from "./edgeTts.js";
+import localDevice, { fetchLocalDeviceVoices } from "./localDevice.js";
+import elevenlabs, { fetchElevenLabsVoices } from "./elevenlabs.js";
+import openai from "./openai.js";
+import openrouter from "./openrouter.js";
+import gemini, { fetchGeminiVoices } from "./gemini.js";
+import xiaomiMimo from "./xiaomi-mimo.js";
+import selfhostedTts from "./selfhostedTts.js";
+import { FORMAT_HANDLERS } from "./genericFormats.js";
+import { parseModelVoice } from "./_base.js";
+
+// Special providers with custom synthesize() logic
+const SPECIAL_ADAPTERS = {
+  "google-tts": googleTts,
+  "edge-tts": edgeTts,
+  "local-device": localDevice,
+  elevenlabs,
+  openai,
+  openrouter,
+  gemini,
+  "xiaomi-mimo": xiaomiMimo,
+  "selfhosted-tts": selfhostedTts,
+};
+
+export function getTtsAdapter(provider) {
+  return SPECIAL_ADAPTERS[provider] || null;
+}
+
+// Generic config-driven dispatcher (uses ttsConfig.format)
+export async function synthesizeViaConfig(provider, text, model, credentials) {
+  const { AI_PROVIDERS } = await import("@/shared/constants/providers");
+  const cfg = AI_PROVIDERS[provider]?.ttsConfig;
+  if (!cfg) return null;
+  const handler = FORMAT_HANDLERS[cfg.format];
+  if (!handler) return null;
+  const apiKey = credentials?.apiKey;
+  if (cfg.authType !== "none" && !apiKey) throw new Error(`${provider} API key required`);
+  const { PROVIDER_MODELS } = await import("open-sse/config/providerModels.js");
+  const ttsModels = (PROVIDER_MODELS[provider] || []).filter((m) => (m.kind || m.type) === "tts");
+  const defaultModel = ttsModels[0]?.id || "";
+  const { modelId, voiceId } = parseModelVoice(model, defaultModel, "", ttsModels);
+  return handler({ baseUrl: cfg.baseUrl, apiKey, text, modelId, voiceId });
+}
+
+// Voice fetchers (used by /api/media-providers/tts/voices route)
+export const VOICE_FETCHERS = {
+  "edge-tts": fetchEdgeTtsVoices,
+  "local-device": fetchLocalDeviceVoices,
+  elevenlabs: fetchElevenLabsVoices,
+  gemini: fetchGeminiVoices,
+};
+
+// Re-export for backward compat
+export { fetchEdgeTtsVoices, fetchLocalDeviceVoices, fetchElevenLabsVoices, fetchGeminiVoices };

package/open-sse/handlers/ttsProviders/localDevice.js

@@ -0,0 +1,98 @@
+// Local device TTS — macOS `say` + Windows SAPI + ffmpeg
+import { execFile } from "node:child_process";
+import { promisify } from "node:util";
+import { mkdtemp, readFile, rm } from "node:fs/promises";
+import { tmpdir } from "node:os";
+import { join } from "node:path";
+
+const execFileAsync = promisify(execFile);
+
+let _voicesCache = null;
+
+async function fetchVoicesMac() {
+  const { stdout } = await execFileAsync("say", ["-v", "?"]);
+  const voices = [];
+  for (const line of stdout.split("\n")) {
+    const m = line.match(/^([^\s].*?)\s{2,}([a-z]{2}_[A-Z]{2})/);
+    if (!m) continue;
+    const name = m[1].trim();
+    const locale = m[2].trim();
+    const lang = locale.split("_")[0];
+    const country = locale.split("_")[1];
+    voices.push({ id: name, name, locale, lang, country, gender: "" });
+  }
+  return voices;
+}
+
+async function fetchVoicesWin() {
+  const script = [
+    "Add-Type -AssemblyName System.Speech;",
+    "$s = New-Object System.Speech.Synthesis.SpeechSynthesizer;",
+    "$s.GetInstalledVoices() | ForEach-Object { $v = $_.VoiceInfo;",
+    "[PSCustomObject]@{ Name=$v.Name; Culture=$v.Culture.Name; Gender=$v.Gender } }",
+    "| ConvertTo-Json -Compress",
+  ].join(" ");
+  const { stdout } = await execFileAsync(
+    "powershell.exe",
+    ["-NoProfile", "-NonInteractive", "-WindowStyle", "Hidden", "-Command", script],
+    { windowsHide: true },
+  );
+  const raw = JSON.parse(stdout.trim() || "[]");
+  const list = Array.isArray(raw) ? raw : [raw];
+  return list.map((v) => {
+    const culture = v.Culture || "en-US";
+    const [lang, country = ""] = culture.split("-");
+    const genderMap = { 1: "Male", 2: "Female", Male: "Male", Female: "Female" };
+    return {
+      id: v.Name,
+      name: v.Name,
+      locale: culture.replace("-", "_"),
+      lang,
+      country,
+      gender: genderMap[v.Gender] || "",
+    };
+  });
+}
+
+export async function fetchLocalDeviceVoices() {
+  if (_voicesCache) return _voicesCache;
+  try {
+    const voices = process.platform === "win32" ? await fetchVoicesWin() : await fetchVoicesMac();
+    _voicesCache = voices;
+    return voices;
+  } catch {
+    return [];
+  }
+}
+
+async function synthesizeMacOrWin(text, voiceId) {
+  const dir = await mkdtemp(join(tmpdir(), "tts-"));
+  const aiffPath = join(dir, "out.aiff");
+  const mp3Path = join(dir, "out.mp3");
+  try {
+    const args = voiceId ? ["-v", voiceId, "-o", aiffPath, text] : ["-o", aiffPath, text];
+    await execFileAsync("say", args);
+    await execFileAsync("ffmpeg", [
+      "-y",
+      "-i",
+      aiffPath,
+      "-codec:a",
+      "libmp3lame",
+      "-qscale:a",
+      "4",
+      mp3Path,
+    ]);
+    const buf = await readFile(mp3Path);
+    return buf.toString("base64");
+  } finally {
+    await rm(dir, { recursive: true, force: true });
+  }
+}
+
+export default {
+  noAuth: true,
+  async synthesize(text, model) {
+    const base64 = await synthesizeMacOrWin(text, model);
+    return { base64, format: "mp3" };
+  },
+};

package/open-sse/handlers/ttsProviders/minimax.js

@@ -0,0 +1,63 @@
+import { Buffer } from "node:buffer";
+
+function hexToBase64(audioHex) {
+  const clean = typeof audioHex === "string" ? audioHex.trim() : "";
+  if (!clean) throw new Error("MiniMax TTS returned no audio");
+  if (clean.length % 2 !== 0 || !/^[0-9a-f]+$/i.test(clean)) {
+    throw new Error("MiniMax TTS returned invalid audio");
+  }
+  return Buffer.from(clean, "hex").toString("base64");
+}
+
+// MiniMax T2A HTTP: returns hex-encoded audio in non-streaming mode.
+export default async function minimaxTts({ baseUrl, apiKey, text, modelId, voiceId }) {
+  const res = await fetch(baseUrl, {
+    method: "POST",
+    headers: { "Content-Type": "application/json", Authorization: `Bearer ${apiKey}` },
+    body: JSON.stringify({
+      model: modelId || "speech-2.8-hd",
+      text,
+      stream: false,
+      language_boost: "auto",
+      output_format: "hex",
+      voice_setting: {
+        voice_id: voiceId || "English_expressive_narrator",
+        speed: 1,
+        vol: 1,
+        pitch: 0,
+      },
+      audio_setting: {
+        sample_rate: 32000,
+        bitrate: 128000,
+        format: "mp3",
+        channel: 1,
+      },
+    }),
+  });
+
+  const rawText = await res.text();
+  let data = {};
+  if (rawText) {
+    try {
+      data = JSON.parse(rawText);
+    } catch {
+      data = {};
+    }
+  }
+
+  const baseResp = data.base_resp || data.baseResp || {};
+  const statusCode = Number(baseResp.status_code ?? baseResp.statusCode ?? 0);
+  const statusMessage = baseResp.status_msg || baseResp.statusMsg || data.message || "";
+
+  if (!res.ok) {
+    throw new Error(statusMessage || rawText || `MiniMax TTS error (${res.status})`);
+  }
+  if (statusCode !== 0) {
+    throw new Error(statusMessage || "MiniMax TTS upstream error");
+  }
+
+  return {
+    base64: hexToBase64(data.data?.audio),
+    format: data.extra_info?.audio_format || data.extraInfo?.audioFormat || "mp3",
+  };
+}

package/open-sse/handlers/ttsProviders/openai.js

@@ -0,0 +1,36 @@
+// OpenAI TTS — model format: "tts-model/voice"
+import { Buffer } from "node:buffer";
+import { PROVIDER_MEDIA } from "../../providers/index.js";
+
+const DEFAULT_TTS_MODEL = PROVIDER_MEDIA["openai"]?.ttsConfig?.defaultModel;
+
+export default {
+  async synthesize(text, model, credentials) {
+    if (!credentials?.apiKey) throw new Error("No OpenAI API key configured");
+
+    let ttsModel = DEFAULT_TTS_MODEL;
+    let voice = "alloy";
+    if (model && model.includes("/")) {
+      const parts = model.split("/");
+      if (parts.length === 2) [ttsModel, voice] = parts;
+    } else if (model) {
+      voice = model;
+    }
+
+    const baseUrl = (credentials.baseUrl || "https://api.openai.com").replace(/\/+$/, "");
+    const res = await fetch(`${baseUrl}/v1/audio/speech`, {
+      method: "POST",
+      headers: {
+        "Content-Type": "application/json",
+        Authorization: `Bearer ${credentials.apiKey}`,
+      },
+      body: JSON.stringify({ model: ttsModel, voice, input: text }),
+    });
+    if (!res.ok) {
+      const err = await res.json().catch(() => ({}));
+      throw new Error(err?.error?.message || `OpenAI TTS failed: ${res.status}`);
+    }
+    const buf = await res.arrayBuffer();
+    return { base64: Buffer.from(buf).toString("base64"), format: "mp3" };
+  },
+};