微信小程序

前提条件

• 已开通智能语音交互服务，并在智能语音交互控制台创建项目、获取 Appkey。

• 已准备微信小程序工程，并在小程序后台配置合法服务器域名：request 域名填写提供 Token 的业务服务端 HTTPS 域名，socket 域名填写 wss://nls-gateway-cn-shanghai.aliyuncs.com。

服务协议、支持的音色和音频参数，请参见语音合成接口说明。

下载安装

从 GitHub 仓库下载 SDK，或直接下载alibabacloud-nls-wx-sdk-master.zip。将 utils 目录放入小程序工程，通过 require("../../utils/tts") 导入 SpeechSynthesizer；相对路径按页面位置调整。

获取 Token

由业务服务端获取 NLS Token，再通过经过身份校验的 HTTPS 接口向小程序提供 Token。获取方法请参见获取 Token。

AccessKey ID 和 AccessKey Secret 只保存在服务端，不写入小程序代码或本地缓存。服务端可根据 Token 的过期时间进行缓存，并在过期前重新获取，避免频繁请求 Token 接口。

代码示例

示例输入文本后合成 WAV 音频，待合成完成后播放。合成和播放期间禁止重复启动，停止操作或页面卸载时关闭连接、销毁播放器并删除本次任务的临时文件。合成超时设置为 60 秒，可按业务需要调整，不是服务端时长限制。

替换 APPKEY 和 TOKEN_URL，并在 getToken() 中补充业务服务端要求的身份认证信息。TOKEN_URL 是自建接口，不是 NLS OpenAPI；示例约定接口返回 token（字符串）和 expireTime（Unix 时间戳，单位为秒），且 Token 剩余有效时间超过 60 秒。

在 app.json 中注册 pages/tts/tts 页面。以下两个文件共同组成页面。

pages/tts/tts.js

const SpeechSynthesizer = require("../../utils/tts");
const APPKEY = "YOUR_APPKEY";
const TOKEN_URL = "https://example.com/api/nls-token";
const URL = "wss://nls-gateway-cn-shanghai.aliyuncs.com/ws/v1";
const SYNTHESIS_TIMEOUT_MS = 60000;
const fs = wx.getFileSystemManager();
let sequence = 0;

function getToken() {
  return new Promise((resolve, reject) => {
    wx.request({
      url: TOKEN_URL,
      method: "GET",
      timeout: 10000,
      // Add the authentication required by your application server.
      success(res) {
        const value = res.data;
        if (res.statusCode === 200 && value &&
            typeof value.token === "string" && value.token &&
            Number(value.expireTime) > Date.now() / 1000 + 60) {
          resolve(value.token);
        } else {
          reject(new Error("Unable to obtain a valid NLS token."));
        }
      },
      fail() { reject(new Error("Token request failed.")); }
    });
  });
}

Page({
  data: { text: "Hello, welcome to speech synthesis.", busy: false, error: "" },

  onLoad() {
    this.alive = true;
    this.task = null;
  },

  textInput(e) { this.setData({ text: e.detail.value }); },

  async onTtsStart() {
    if (!this.alive || this.task) return;
    const text = this.data.text.trim();
    if (!text) {
      this.setData({ error: "Enter text to synthesize." });
      return;
    }
    const task = {
      path: `${wx.env.USER_DATA_PATH}/tts-${Date.now()}-${++sequence}.wav`,
      phase: "token", bytes: 0, fileCreated: false, disposed: false
    };
    this.task = task;
    this.setData({ busy: true, error: "" });

    try {
      const token = await getToken();
      if (task.disposed) return;
      fs.writeFileSync(task.path, new ArrayBuffer(0));
      task.fileCreated = true;
      task.tts = new SpeechSynthesizer({ url: URL, appkey: APPKEY, token });
      task.phase = "synthesis";

      // Connection closure alone does not settle the SDK's start Promise.
      const interrupted = new Promise((resolve, reject) => {
        task.reject = reject;
        task.timer = setTimeout(() => {
          reject(new Error("Speech synthesis timed out."));
        }, SYNTHESIS_TIMEOUT_MS);
      });
      task.tts.on("data", data => {
        if (task.disposed) return;
        try {
          fs.appendFileSync(task.path, data);
          task.bytes += data.byteLength;
        } catch (_) {
          task.reject(new Error("Unable to save audio."));
        }
      });
      task.tts.on("completed", () => { task.phase = "complete"; });
      task.tts.on("closed", () => {
        if (!task.disposed && task.phase === "synthesis") {
          task.reject(new Error("Connection closed before synthesis completed."));
        }
      });
      task.tts.on("failed", message => {
        // Log only diagnostic fields, not credentials or the full response.
        try {
          const header = JSON.parse(message).header;
          console.error("Synthesis failed", header.status, header.task_id);
        } catch (_) { console.error("Synthesis failed"); }
      });

      const params = task.tts.defaultStartParams("aixia");
      params.text = text;
      params.format = "wav";
      await Promise.race([task.tts.start(params), interrupted]);
      clearTimeout(task.timer);
      if (task.disposed) return;
      if (!task.bytes) throw new Error("No audio was returned.");

      task.phase = "playback";
      const audio = wx.createInnerAudioContext();
      task.audio = audio;
      audio.onEnded(() => this.dispose(task));
      audio.onError(() => this.dispose(task, "Unable to play audio."));
      audio.src = task.path;
      audio.play();
    } catch (_) {
      if (!task.disposed) {
        this.dispose(task, "Synthesis failed. Check the token, network, and audio settings.");
      }
    }
  },

  dispose(task, error = "") {
    if (!task || task.disposed) return;
    task.disposed = true;
    clearTimeout(task.timer);
    if (task.reject) task.reject(new Error("Task cancelled."));
    if (task.tts) task.tts.shutdown();
    if (task.audio) {
      task.audio.stop();
      task.audio.destroy();
    }
    if (task.fileCreated) {
      fs.unlink({
        filePath: task.path,
        fail() { console.warn("Unable to remove the temporary audio file."); }
      });
    }
    if (this.task === task) {
      this.task = null;
      if (this.alive) this.setData({ busy: false, error });
    }
  },

  onStop() { this.dispose(this.task); },

  onUnload() {
    this.alive = false;
    this.dispose(this.task);
  }
});

pages/tts/tts.wxml

<view>
  <textarea bindinput="textInput" value="{{text}}" maxlength="300" disabled="{{busy}}" />
  <button bindtap="onTtsStart" disabled="{{busy}}">合成并播放</button>
  <button bindtap="onStop" disabled="{{!busy}}">停止</button>
  <text>{{error}}</text>
</view>

API 参考

const params = tts.defaultStartParams("aixia");
params.text = "Hello, welcome to speech synthesis.";
params.enable_subtitle = true;

const params = {
  text: "Hello, welcome to speech synthesis.",
  voice: "aixia",
  format: "wav",
  sample_rate: 16000,
  volume: 50,
  speech_rate: 0,
  pitch_rate: 0,
  enable_subtitle: true
};