CorpusAI/service/speech_to_text.go

package service

import (
	"bytes"
	"crypto/md5"
	"encoding/base64"
	"encoding/hex"
	"encoding/json"
	"errors"
	"fmt"
	"io/ioutil"
	"mime/multipart"
	"net/http"
	"net/url"
	"strings"
	"time"
)

// Xfyun 讯飞云调用对象
type Xfyun struct {
	URL string
	ID  string
	KEY string
}

// XfyunResult 讯飞云返回结果
type XfyunResult struct {
	Code string `json:"code"`
	Data string `json:"data"`
	Desc string `json:"desc"`
	Sid  string `json:"sid"`
}

// XfyunParam 讯飞云请求参数
type XfyunParam struct {
	Aue        string `json:"aue"`
	EngineType string `json:"engine_type"`
}

// SuccessCode 讯飞云成功代码
const SuccessCode = "success"

// SpeechToText speech to text input File
func SpeechToText(file multipart.File, lang string, rateStr string) (string, error) {
	xfyun := Xfyun{
		URL: "http://api.xfyun.cn/v1/service/v1/iat",
		ID:  "5bbaeda1",
		KEY: "25c68c82b45309d35a7f14069107746b",
	}
	wav, err := ioutil.ReadAll(file)

	if err != nil {
		return "", err
	}
	//
	base64Audio := base64.StdEncoding.EncodeToString(wav)
	u := url.Values{}
	u.Set("audio", base64Audio)
	body := u.Encode()
	xfyunParam := XfyunParam{
		Aue:        "raw",
		EngineType: "sms",
	}

	if lang != "zh" {
		xfyunParam.EngineType += "-en"
	}
	if rateStr == "8" {
		xfyunParam.EngineType += "8k"
	} else {
		xfyunParam.EngineType += "16k"
	}
	xfyunParamString, _ := json.Marshal(xfyunParam)

	fmt.Println(string(xfyunParamString))

	xParam := base64.StdEncoding.EncodeToString(xfyunParamString)
	xTime := time.Now().Unix()

	h := md5.New()
	h.Write([]byte(fmt.Sprintf("%s%d%s", xfyun.KEY, xTime, xParam)))
	xChecksum := hex.EncodeToString(h.Sum(nil))

	client := &http.Client{}

	req, _ := http.NewRequest("POST", xfyun.URL, bytes.NewBuffer([]byte(body)))

	req.Header.Set("Accept-Encoding", "identity")
	req.Header.Set("Content-Type", "application/x-www-form-urlencoded")
	req.Header.Add("X-Appid", xfyun.ID)
	req.Header.Add("X-CurTime", fmt.Sprintf("%d", xTime))
	req.Header.Add("X-Param", xParam)
	req.Header.Add("X-CheckSum", xChecksum)

	resp, err := client.Do(req)
	if err != nil {
		return "", nil
	}

	html, err := ioutil.ReadAll(resp.Body)
	if err != nil {
		return "", err
	}

	xfyunResult := XfyunResult{}
	err = json.Unmarshal(html, &xfyunResult)
	if err != nil {
		return "", nil
	}
	var text string
	if xfyunResult.Desc == SuccessCode {
		text = xfyunResult.Data
	} else {
		err = errors.New(xfyunResult.Desc)
		return "", err
	}

	return strings.TrimSpace(text), nil
}