微软文本转语音实测记录

微软文本转语音

接口地址

地址：http://www.mysqlschool.cn/SpeekText/index.php

提交方式：post/get  推荐post

单段落例子：http://www.mysqlschool.cn/SpeekText/index.php

多段落例子：http://www.mysqlschool.cn/SpeekText/SpeekText.html

生成失败无需刷新页面重新点击生成语音即可

参数讲解

voice_name 可选项

zh-CN-XiaoxiaoNeural > Xiaoxiao (Neural) - 晓晓 

zh-CN-YunyangNeural > Yunyang (Neural) - 云扬 

zh-CN-XiaochenNeural > Xiaochen (Neural) - 晓辰

zh-CN-XiaohanNeural > Xiaohan (Neural) - 晓涵

zh-CN-XiaomoNeural > Xiaomo (Neural) - 晓墨

zh-CN-XiaoqiuNeural > Xiaoqiu (Neural) - 晓秋

zh-CN-XiaoruiNeural > Xiaorui (Neural) - 晓睿

zh-CN-XiaoshuangNeural > Xiaoshuang (Neural) - 晓双

zh-CN-XiaoxuanNeural > Xiaoxuan (Neural) - 晓萱

zh-CN-XiaoyanNeural > Xiaoyan (Neural) - 晓颜

zh-CN-XiaoyouNeural > Xiaoyou (Neural) - 晓悠

zh-CN-YunxiNeural > Yunxi (Neural) - 云希

zh-CN-YunyeNeural > Yunye (Neural) - 云野

role 可选项

Style 可选项

其他参数请看《通过语音合成标记语言 (SSML) 改善合成》

预生成的神经语音

预生成的神经网络声音是从使用 24 khz 采样率的样本创建的。 合成时，所有声音都可以使用其他采样率进行上采样或下采样。

下表列出了每种语言支持的预生成的神经网络声音。

这里只写了中文的各种参数配置，全部配置可看《语音服务的语言和语音支持》

自己写的php接口源码

GO源码

其他语言源码可看《Cognitive-Speech-TTS》

package main

import (
	"bytes"
	"io"
	"io/ioutil"
	"log"
	"net/http"
	"os"
	"time"
)

// auth token
// var token = ""

func GetToken(auth string, key string) string {
	postData := []byte("")

	req, err := http.NewRequest("POST", auth, bytes.NewBuffer(postData))
	if err != nil {
		log.Printf("New request with error %s\n", err)
	}

	token := ""
	req.Header.Set("Ocp-Apim-Subscription-Key", key)
	response, err := http.DefaultClient.Do(req)
	if err != nil {
		log.Printf("Do request failed with error %s\n", err)
	} else {
		defer response.Body.Close()
		data, _ := ioutil.ReadAll(response.Body)
		token = string(data)
		log.Println("token is", token)
	}

	return token
}

func GetAudioBytes(client *http.Client, endpoint string, authToken string, ssml string) {
	start := time.Now()
	postData := []byte(ssml)
	req, err := http.NewRequest("POST", endpoint, bytes.NewBuffer(postData))
	if err != nil {
		log.Printf("New request with error %s\n", err)
		return
	}

	req.Header.Set("content-type", "application/ssml+xml")
	req.Header.Set("X-Microsoft-OutputFormat", "riff-24khz-16bit-mono-pcm")
	req.Header.Set("Authorization", "Bearer "+authToken)
	req.Header.Set("User-Agent", "GoClient")
	response, err := client.Do(req)
	if err != nil {
		log.Printf("Do request failed with error %s\n", err)
	} else {
		defer response.Body.Close()
		if response.StatusCode == 200 {
			data, _ := ioutil.ReadAll(response.Body)
			log.Println("audio byte len is", len(data))
			ioutil.WriteFile("output.mp3", data, 0644)
		} else {
			log.Println("status code is", response.StatusCode)
		}
	}

	elapsed := time.Since(start)
	log.Printf("GetAudioBytes took %s", elapsed)
}

func main() {

	f, err := os.OpenFile("run.log", os.O_RDWR|os.O_CREATE|os.O_APPEND, 0666)
	if err != nil {
		log.Fatalf("error opening file: %v", err)
		return
	}
	defer f.Close()
	wrt := io.MultiWriter(os.Stdout, f)
	log.SetOutput(wrt)
	log.Println("Starting the application...")

	endpoint := "https://注册时写的位置.tts.speech.microsoft.com/cognitiveservices/v1"
	auth := "https://注册时写的位置.api.cognitive.microsoft.com/sts/v1.0/issueToken"
	key := "你的key"

	token := GetToken(auth, key)
	ticker := time.NewTicker(300 * time.Second)
	go func() {
		for range ticker.C {
			log.Println("Refreshing token")
			t := GetToken(auth, key)
			if len(t) != 0 {
				token = t
			}
		}
	}()

	tr := &http.Transport{
		MaxIdleConns:        100,
		MaxIdleConnsPerHost: 20,
	}

	client := &http.Client{
		Transport: tr,
	}

	GetAudioBytes(client, endpoint, token, "<speak version='1.0' xml:lang='zh-CN'><voice name='zh-CN-YunyeNeural'>哈哈哈，终于成功了!</voice></speak>")

}

这花开吗SVIP

打赏 收藏 海报 链接