Merge pull request #367 from Calcium-Ion/audio

feat: support cloudflare tts

relay/channel/adapter.go

@@ -10,12 +10,13 @@ import (
 
 type Adaptor interface {
 	// Init IsStream bool
-	Init(info *relaycommon.RelayInfo, request dto.GeneralOpenAIRequest)
-	InitRerank(info *relaycommon.RelayInfo, request dto.RerankRequest)
+	Init(info *relaycommon.RelayInfo)
 	GetRequestURL(info *relaycommon.RelayInfo) (string, error)
 	SetupRequestHeader(c *gin.Context, req *http.Request, info *relaycommon.RelayInfo) error
 	ConvertRequest(c *gin.Context, info *relaycommon.RelayInfo, request *dto.GeneralOpenAIRequest) (any, error)
 	ConvertRerankRequest(c *gin.Context, relayMode int, request dto.RerankRequest) (any, error)
+	ConvertAudioRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.AudioRequest) (io.Reader, error)
+	ConvertImageRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.ImageRequest) (any, error)
 	DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (*http.Response, error)
 	DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage *dto.Usage, err *dto.OpenAIErrorWithStatusCode)
 	GetModelList() []string

relay/channel/ali/adaptor.go

@@ -15,11 +15,17 @@ import (
 type Adaptor struct {
 }
 
-func (a *Adaptor) InitRerank(info *relaycommon.RelayInfo, request dto.RerankRequest) {
+func (a *Adaptor) ConvertAudioRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.AudioRequest) (io.Reader, error) {
+	//TODO implement me
+	return nil, errors.New("not implemented")
 }
 
-func (a *Adaptor) Init(info *relaycommon.RelayInfo, request dto.GeneralOpenAIRequest) {
+func (a *Adaptor) ConvertImageRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.ImageRequest) (any, error) {
+	//TODO implement me
+	return nil, errors.New("not implemented")
+}
 
+func (a *Adaptor) Init(info *relaycommon.RelayInfo) {
 }
 
 func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {

relay/channel/api_request.go

@@ -7,14 +7,19 @@ import (
 	"io"
 	"net/http"
 	"one-api/relay/common"
+	"one-api/relay/constant"
 	"one-api/service"
 )
 
 func SetupApiRequestHeader(info *common.RelayInfo, c *gin.Context, req *http.Request) {
-	req.Header.Set("Content-Type", c.Request.Header.Get("Content-Type"))
-	req.Header.Set("Accept", c.Request.Header.Get("Accept"))
-	if info.IsStream && c.Request.Header.Get("Accept") == "" {
-		req.Header.Set("Accept", "text/event-stream")
+	if info.RelayMode == constant.RelayModeAudioTranscription || info.RelayMode == constant.RelayModeAudioTranslation {
+		// multipart/form-data
+	} else {
+		req.Header.Set("Content-Type", c.Request.Header.Get("Content-Type"))
+		req.Header.Set("Accept", c.Request.Header.Get("Accept"))
+		if info.IsStream && c.Request.Header.Get("Accept") == "" {
+			req.Header.Set("Accept", "text/event-stream")
+		}
 	}
 }
 
@@ -38,6 +43,29 @@ func DoApiRequest(a Adaptor, c *gin.Context, info *common.RelayInfo, requestBody
 	return resp, nil
 }
 
+func DoFormRequest(a Adaptor, c *gin.Context, info *common.RelayInfo, requestBody io.Reader) (*http.Response, error) {
+	fullRequestURL, err := a.GetRequestURL(info)
+	if err != nil {
+		return nil, fmt.Errorf("get request url failed: %w", err)
+	}
+	req, err := http.NewRequest(c.Request.Method, fullRequestURL, requestBody)
+	if err != nil {
+		return nil, fmt.Errorf("new request failed: %w", err)
+	}
+	// set form data
+	req.Header.Set("Content-Type", c.Request.Header.Get("Content-Type"))
+
+	err = a.SetupRequestHeader(c, req, info)
+	if err != nil {
+		return nil, fmt.Errorf("setup request header failed: %w", err)
+	}
+	resp, err := doRequest(c, req)
+	if err != nil {
+		return nil, fmt.Errorf("do request failed: %w", err)
+	}
+	return resp, nil
+}
+
 func doRequest(c *gin.Context, req *http.Request) (*http.Response, error) {
 	resp, err := service.GetHttpClient().Do(req)
 	if err != nil {

relay/channel/aws/adaptor.go

@@ -20,12 +20,17 @@ type Adaptor struct {
 	RequestMode int
 }
 
-func (a *Adaptor) InitRerank(info *relaycommon.RelayInfo, request dto.RerankRequest) {
+func (a *Adaptor) ConvertAudioRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.AudioRequest) (io.Reader, error) {
 	//TODO implement me
-
+	return nil, errors.New("not implemented")
 }
 
-func (a *Adaptor) Init(info *relaycommon.RelayInfo, request dto.GeneralOpenAIRequest) {
+func (a *Adaptor) ConvertImageRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.ImageRequest) (any, error) {
+	//TODO implement me
+	return nil, errors.New("not implemented")
+}
+
+func (a *Adaptor) Init(info *relaycommon.RelayInfo) {
 	if strings.HasPrefix(info.UpstreamModelName, "claude-3") {
 		a.RequestMode = RequestModeMessage
 	} else {

relay/channel/baidu/adaptor.go

@@ -16,12 +16,17 @@ import (
 type Adaptor struct {
 }
 
-func (a *Adaptor) InitRerank(info *relaycommon.RelayInfo, request dto.RerankRequest) {
+func (a *Adaptor) ConvertAudioRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.AudioRequest) (io.Reader, error) {
 	//TODO implement me
-
+	return nil, errors.New("not implemented")
 }
 
-func (a *Adaptor) Init(info *relaycommon.RelayInfo, request dto.GeneralOpenAIRequest) {
+func (a *Adaptor) ConvertImageRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.ImageRequest) (any, error) {
+	//TODO implement me
+	return nil, errors.New("not implemented")
+}
+
+func (a *Adaptor) Init(info *relaycommon.RelayInfo) {
 
 }
 

relay/channel/claude/adaptor.go

@@ -21,12 +21,17 @@ type Adaptor struct {
 	RequestMode int
 }
 
-func (a *Adaptor) InitRerank(info *relaycommon.RelayInfo, request dto.RerankRequest) {
+func (a *Adaptor) ConvertAudioRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.AudioRequest) (io.Reader, error) {
 	//TODO implement me
-
+	return nil, errors.New("not implemented")
 }
 
-func (a *Adaptor) Init(info *relaycommon.RelayInfo, request dto.GeneralOpenAIRequest) {
+func (a *Adaptor) ConvertImageRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.ImageRequest) (any, error) {
+	//TODO implement me
+	return nil, errors.New("not implemented")
+}
+
+func (a *Adaptor) Init(info *relaycommon.RelayInfo) {
 	if strings.HasPrefix(info.UpstreamModelName, "claude-3") {
 		a.RequestMode = RequestModeMessage
 	} else {

relay/channel/cloudflare/adaptor.go

@@ -1,6 +1,7 @@
 package cloudflare
 
 import (
+	"bytes"
 	"errors"
 	"fmt"
 	"github.com/gin-gonic/gin"
@@ -15,10 +16,7 @@ import (
 type Adaptor struct {
 }
 
-func (a *Adaptor) InitRerank(info *relaycommon.RelayInfo, request dto.RerankRequest) {
-}
-
-func (a *Adaptor) Init(info *relaycommon.RelayInfo, request dto.GeneralOpenAIRequest) {
+func (a *Adaptor) Init(info *relaycommon.RelayInfo) {
 }
 
 func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {
@@ -58,11 +56,42 @@ func (a *Adaptor) ConvertRerankRequest(c *gin.Context, relayMode int, request dt
 	return request, nil
 }
 
+func (a *Adaptor) ConvertAudioRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.AudioRequest) (io.Reader, error) {
+	// 添加文件字段
+	file, _, err := c.Request.FormFile("file")
+	if err != nil {
+		return nil, errors.New("file is required")
+	}
+	defer file.Close()
+	// 打开临时文件用于保存上传的文件内容
+	requestBody := &bytes.Buffer{}
+
+	// 将上传的文件内容复制到临时文件
+	if _, err := io.Copy(requestBody, file); err != nil {
+		return nil, err
+	}
+	return requestBody, nil
+}
+
+func (a *Adaptor) ConvertImageRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.ImageRequest) (any, error) {
+	//TODO implement me
+	return nil, errors.New("not implemented")
+}
+
 func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage *dto.Usage, err *dto.OpenAIErrorWithStatusCode) {
-	if info.IsStream {
-		err, usage = cfStreamHandler(c, resp, info)
-	} else {
-		err, usage = cfHandler(c, resp, info)
+	switch info.RelayMode {
+	case constant.RelayModeEmbeddings:
+		fallthrough
+	case constant.RelayModeChatCompletions:
+		if info.IsStream {
+			err, usage = cfStreamHandler(c, resp, info)
+		} else {
+			err, usage = cfHandler(c, resp, info)
+		}
+	case constant.RelayModeAudioTranslation:
+		fallthrough
+	case constant.RelayModeAudioTranscription:
+		err, usage = cfSTTHandler(c, resp, info)
 	}
 	return
 }

relay/channel/cloudflare/dto.go

@@ -11,3 +11,11 @@ type CfRequest struct {
 	Stream      bool          `json:"stream,omitempty"`
 	Temperature float64       `json:"temperature,omitempty"`
 }
+
+type CfAudioResponse struct {
+	Result CfSTTResult `json:"result"`
+}
+
+type CfSTTResult struct {
+	Text string `json:"text"`
+}

relay/channel/cloudflare/relay_cloudflare.go

@@ -119,3 +119,38 @@ func cfHandler(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo)
 	_, _ = c.Writer.Write(jsonResponse)
 	return nil, usage
 }
+
+func cfSTTHandler(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (*dto.OpenAIErrorWithStatusCode, *dto.Usage) {
+	var cfResp CfAudioResponse
+	responseBody, err := io.ReadAll(resp.Body)
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "read_response_body_failed", http.StatusInternalServerError), nil
+	}
+	err = resp.Body.Close()
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), nil
+	}
+	err = json.Unmarshal(responseBody, &cfResp)
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "unmarshal_response_body_failed", http.StatusInternalServerError), nil
+	}
+
+	audioResp := &dto.AudioResponse{
+		Text: cfResp.Result.Text,
+	}
+
+	jsonResponse, err := json.Marshal(audioResp)
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "marshal_response_body_failed", http.StatusInternalServerError), nil
+	}
+	c.Writer.Header().Set("Content-Type", "application/json")
+	c.Writer.WriteHeader(resp.StatusCode)
+	_, _ = c.Writer.Write(jsonResponse)
+
+	usage := &dto.Usage{}
+	usage.PromptTokens = info.PromptTokens
+	usage.CompletionTokens, _ = service.CountTokenText(cfResp.Result.Text, info.UpstreamModelName)
+	usage.TotalTokens = usage.PromptTokens + usage.CompletionTokens
+
+	return nil, usage
+}

relay/channel/cohere/adaptor.go

@@ -1,6 +1,7 @@
 package cohere
 
 import (
+	"errors"
 	"fmt"
 	"github.com/gin-gonic/gin"
 	"io"
@@ -14,10 +15,17 @@ import (
 type Adaptor struct {
 }
 
-func (a *Adaptor) InitRerank(info *relaycommon.RelayInfo, request dto.RerankRequest) {
+func (a *Adaptor) ConvertAudioRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.AudioRequest) (io.Reader, error) {
+	//TODO implement me
+	return nil, errors.New("not implemented")
 }
 
-func (a *Adaptor) Init(info *relaycommon.RelayInfo, request dto.GeneralOpenAIRequest) {
+func (a *Adaptor) ConvertImageRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.ImageRequest) (any, error) {
+	//TODO implement me
+	return nil, errors.New("not implemented")
+}
+
+func (a *Adaptor) Init(info *relaycommon.RelayInfo) {
 }
 
 func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {

relay/channel/dify/adaptor.go

@@ -14,12 +14,17 @@ import (
 type Adaptor struct {
 }
 
-func (a *Adaptor) InitRerank(info *relaycommon.RelayInfo, request dto.RerankRequest) {
+func (a *Adaptor) ConvertAudioRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.AudioRequest) (io.Reader, error) {
 	//TODO implement me
-
+	return nil, errors.New("not implemented")
 }
 
-func (a *Adaptor) Init(info *relaycommon.RelayInfo, request dto.GeneralOpenAIRequest) {
+func (a *Adaptor) ConvertImageRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.ImageRequest) (any, error) {
+	//TODO implement me
+	return nil, errors.New("not implemented")
+}
+
+func (a *Adaptor) Init(info *relaycommon.RelayInfo) {
 }
 
 func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {

relay/channel/gemini/adaptor.go

@@ -14,10 +14,17 @@ import (
 type Adaptor struct {
 }
 
-func (a *Adaptor) InitRerank(info *relaycommon.RelayInfo, request dto.RerankRequest) {
+func (a *Adaptor) ConvertAudioRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.AudioRequest) (io.Reader, error) {
+	//TODO implement me
+	return nil, errors.New("not implemented")
 }
 
-func (a *Adaptor) Init(info *relaycommon.RelayInfo, request dto.GeneralOpenAIRequest) {
+func (a *Adaptor) ConvertImageRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.ImageRequest) (any, error) {
+	//TODO implement me
+	return nil, errors.New("not implemented")
+}
+
+func (a *Adaptor) Init(info *relaycommon.RelayInfo) {
 }
 
 // 定义一个映射，存储模型名称和对应的版本

relay/channel/jina/adaptor.go

@@ -15,10 +15,17 @@ import (
 type Adaptor struct {
 }
 
-func (a *Adaptor) InitRerank(info *relaycommon.RelayInfo, request dto.RerankRequest) {
+func (a *Adaptor) ConvertAudioRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.AudioRequest) (io.Reader, error) {
+	//TODO implement me
+	return nil, errors.New("not implemented")
 }
 
-func (a *Adaptor) Init(info *relaycommon.RelayInfo, request dto.GeneralOpenAIRequest) {
+func (a *Adaptor) ConvertImageRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.ImageRequest) (any, error) {
+	//TODO implement me
+	return nil, errors.New("not implemented")
+}
+
+func (a *Adaptor) Init(info *relaycommon.RelayInfo) {
 }
 
 func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {

relay/channel/ollama/adaptor.go

@@ -15,10 +15,17 @@ import (
 type Adaptor struct {
 }
 
-func (a *Adaptor) InitRerank(info *relaycommon.RelayInfo, request dto.RerankRequest) {
+func (a *Adaptor) ConvertAudioRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.AudioRequest) (io.Reader, error) {
+	//TODO implement me
+	return nil, errors.New("not implemented")
 }
 
-func (a *Adaptor) Init(info *relaycommon.RelayInfo, request dto.GeneralOpenAIRequest) {
+func (a *Adaptor) ConvertImageRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.ImageRequest) (any, error) {
+	//TODO implement me
+	return nil, errors.New("not implemented")
+}
+
+func (a *Adaptor) Init(info *relaycommon.RelayInfo) {
 }
 
 func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {

relay/channel/openai/adaptor.go

@@ -1,10 +1,13 @@
 package openai
 
 import (
+	"bytes"
+	"encoding/json"
 	"errors"
 	"fmt"
 	"github.com/gin-gonic/gin"
 	"io"
+	"mime/multipart"
 	"net/http"
 	"one-api/common"
 	"one-api/dto"
@@ -14,21 +17,16 @@ import (
 	"one-api/relay/channel/minimax"
 	"one-api/relay/channel/moonshot"
 	relaycommon "one-api/relay/common"
+	"one-api/relay/constant"
 	"strings"
 )
 
 type Adaptor struct {
-	ChannelType int
+	ChannelType    int
+	ResponseFormat string
 }
 
-func (a *Adaptor) ConvertRerankRequest(c *gin.Context, relayMode int, request dto.RerankRequest) (any, error) {
-	return nil, nil
-}
-
-func (a *Adaptor) InitRerank(info *relaycommon.RelayInfo, request dto.RerankRequest) {
-}
-
-func (a *Adaptor) Init(info *relaycommon.RelayInfo, request dto.GeneralOpenAIRequest) {
+func (a *Adaptor) Init(info *relaycommon.RelayInfo) {
 	a.ChannelType = info.ChannelType
 }
 
@@ -83,15 +81,73 @@ func (a *Adaptor) ConvertRequest(c *gin.Context, info *relaycommon.RelayInfo, re
 	return request, nil
 }
 
+func (a *Adaptor) ConvertRerankRequest(c *gin.Context, relayMode int, request dto.RerankRequest) (any, error) {
+	return nil, errors.New("not implemented")
+}
+
+func (a *Adaptor) ConvertAudioRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.AudioRequest) (io.Reader, error) {
+	a.ResponseFormat = request.ResponseFormat
+	if info.RelayMode == constant.RelayModeAudioSpeech {
+		jsonData, err := json.Marshal(request)
+		if err != nil {
+			return nil, fmt.Errorf("error marshalling object: %w", err)
+		}
+		return bytes.NewReader(jsonData), nil
+	} else {
+		var requestBody bytes.Buffer
+		writer := multipart.NewWriter(&requestBody)
+
+		writer.WriteField("model", request.Model)
+
+		// 添加文件字段
+		file, header, err := c.Request.FormFile("file")
+		if err != nil {
+			return nil, errors.New("file is required")
+		}
+		defer file.Close()
+
+		part, err := writer.CreateFormFile("file", header.Filename)
+		if err != nil {
+			return nil, errors.New("create form file failed")
+		}
+		if _, err := io.Copy(part, file); err != nil {
+			return nil, errors.New("copy file failed")
+		}
+
+		// 关闭 multipart 编写器以设置分界线
+		writer.Close()
+		c.Request.Header.Set("Content-Type", writer.FormDataContentType())
+		return &requestBody, nil
+	}
+}
+
+func (a *Adaptor) ConvertImageRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.ImageRequest) (any, error) {
+	//TODO implement me
+	return nil, errors.New("not implemented")
+}
+
 func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (*http.Response, error) {
-	return channel.DoApiRequest(a, c, info, requestBody)
+	if info.RelayMode == constant.RelayModeAudioTranscription || info.RelayMode == constant.RelayModeAudioTranslation {
+		return channel.DoFormRequest(a, c, info, requestBody)
+	} else {
+		return channel.DoApiRequest(a, c, info, requestBody)
+	}
 }
 
 func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage *dto.Usage, err *dto.OpenAIErrorWithStatusCode) {
-	if info.IsStream {
-		err, usage = OpenaiStreamHandler(c, resp, info)
-	} else {
-		err, usage = OpenaiHandler(c, resp, info.PromptTokens, info.UpstreamModelName)
+	switch info.RelayMode {
+	case constant.RelayModeAudioSpeech:
+		err, usage = OpenaiTTSHandler(c, resp, info)
+	case constant.RelayModeAudioTranslation:
+		fallthrough
+	case constant.RelayModeAudioTranscription:
+		err, usage = OpenaiSTTHandler(c, resp, info, a.ResponseFormat)
+	default:
+		if info.IsStream {
+			err, usage = OpenaiStreamHandler(c, resp, info)
+		} else {
+			err, usage = OpenaiHandler(c, resp, info.PromptTokens, info.UpstreamModelName)
+		}
 	}
 	return
 }

relay/channel/openai/relay-openai.go

@@ -4,6 +4,7 @@ import (
 	"bufio"
 	"bytes"
 	"encoding/json"
+	"fmt"
 	"github.com/gin-gonic/gin"
 	"io"
 	"net/http"
@@ -167,10 +168,7 @@ func OpenaiStreamHandler(c *gin.Context, resp *http.Response, info *relaycommon.
 
 	service.Done(c)
 
-	err := resp.Body.Close()
-	if err != nil {
-		common.LogError(c, "close_response_body_failed: "+err.Error())
-	}
+	resp.Body.Close()
 	return nil, usage
 }
 
@@ -208,11 +206,7 @@ func OpenaiHandler(c *gin.Context, resp *http.Response, promptTokens int, model
 	if err != nil {
 		return service.OpenAIErrorWrapper(err, "copy_response_body_failed", http.StatusInternalServerError), nil
 	}
-	err = resp.Body.Close()
-	if err != nil {
-		return service.OpenAIErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), nil
-	}
-
+	resp.Body.Close()
 	if simpleResponse.Usage.TotalTokens == 0 || (simpleResponse.Usage.PromptTokens == 0 && simpleResponse.Usage.CompletionTokens == 0) {
 		completionTokens := 0
 		for _, choice := range simpleResponse.Choices {
@@ -227,3 +221,134 @@ func OpenaiHandler(c *gin.Context, resp *http.Response, promptTokens int, model
 	}
 	return nil, &simpleResponse.Usage
 }
+
+func OpenaiTTSHandler(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (*dto.OpenAIErrorWithStatusCode, *dto.Usage) {
+	responseBody, err := io.ReadAll(resp.Body)
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "read_response_body_failed", http.StatusInternalServerError), nil
+	}
+	err = resp.Body.Close()
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), nil
+	}
+	// Reset response body
+	resp.Body = io.NopCloser(bytes.NewBuffer(responseBody))
+	// We shouldn't set the header before we parse the response body, because the parse part may fail.
+	// And then we will have to send an error response, but in this case, the header has already been set.
+	// So the httpClient will be confused by the response.
+	// For example, Postman will report error, and we cannot check the response at all.
+	for k, v := range resp.Header {
+		c.Writer.Header().Set(k, v[0])
+	}
+	c.Writer.WriteHeader(resp.StatusCode)
+	_, err = io.Copy(c.Writer, resp.Body)
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "copy_response_body_failed", http.StatusInternalServerError), nil
+	}
+	err = resp.Body.Close()
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), nil
+	}
+
+	usage := &dto.Usage{}
+	usage.PromptTokens = info.PromptTokens
+	usage.TotalTokens = info.PromptTokens
+	return nil, usage
+}
+
+func OpenaiSTTHandler(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo, responseFormat string) (*dto.OpenAIErrorWithStatusCode, *dto.Usage) {
+	var audioResp dto.AudioResponse
+	responseBody, err := io.ReadAll(resp.Body)
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "read_response_body_failed", http.StatusInternalServerError), nil
+	}
+	err = resp.Body.Close()
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), nil
+	}
+	err = json.Unmarshal(responseBody, &audioResp)
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "unmarshal_response_body_failed", http.StatusInternalServerError), nil
+	}
+
+	// Reset response body
+	resp.Body = io.NopCloser(bytes.NewBuffer(responseBody))
+	// We shouldn't set the header before we parse the response body, because the parse part may fail.
+	// And then we will have to send an error response, but in this case, the header has already been set.
+	// So the httpClient will be confused by the response.
+	// For example, Postman will report error, and we cannot check the response at all.
+	for k, v := range resp.Header {
+		c.Writer.Header().Set(k, v[0])
+	}
+	c.Writer.WriteHeader(resp.StatusCode)
+	_, err = io.Copy(c.Writer, resp.Body)
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "copy_response_body_failed", http.StatusInternalServerError), nil
+	}
+	resp.Body.Close()
+
+	var text string
+	switch responseFormat {
+	case "json":
+		text, err = getTextFromJSON(responseBody)
+	case "text":
+		text, err = getTextFromText(responseBody)
+	case "srt":
+		text, err = getTextFromSRT(responseBody)
+	case "verbose_json":
+		text, err = getTextFromVerboseJSON(responseBody)
+	case "vtt":
+		text, err = getTextFromVTT(responseBody)
+	}
+
+	usage := &dto.Usage{}
+	usage.PromptTokens = info.PromptTokens
+	usage.CompletionTokens, _ = service.CountTokenText(text, info.UpstreamModelName)
+	usage.TotalTokens = usage.PromptTokens + usage.CompletionTokens
+	return nil, usage
+}
+
+func getTextFromVTT(body []byte) (string, error) {
+	return getTextFromSRT(body)
+}
+
+func getTextFromVerboseJSON(body []byte) (string, error) {
+	var whisperResponse dto.WhisperVerboseJSONResponse
+	if err := json.Unmarshal(body, &whisperResponse); err != nil {
+		return "", fmt.Errorf("unmarshal_response_body_failed err :%w", err)
+	}
+	return whisperResponse.Text, nil
+}
+
+func getTextFromSRT(body []byte) (string, error) {
+	scanner := bufio.NewScanner(strings.NewReader(string(body)))
+	var builder strings.Builder
+	var textLine bool
+	for scanner.Scan() {
+		line := scanner.Text()
+		if textLine {
+			builder.WriteString(line)
+			textLine = false
+			continue
+		} else if strings.Contains(line, "-->") {
+			textLine = true
+			continue
+		}
+	}
+	if err := scanner.Err(); err != nil {
+		return "", err
+	}
+	return builder.String(), nil
+}
+
+func getTextFromText(body []byte) (string, error) {
+	return strings.TrimSuffix(string(body), "\n"), nil
+}
+
+func getTextFromJSON(body []byte) (string, error) {
+	var whisperResponse dto.AudioResponse
+	if err := json.Unmarshal(body, &whisperResponse); err != nil {
+		return "", fmt.Errorf("unmarshal_response_body_failed err :%w", err)
+	}
+	return whisperResponse.Text, nil
+}

relay/channel/palm/adaptor.go

@@ -15,12 +15,17 @@ import (
 type Adaptor struct {
 }
 
-func (a *Adaptor) InitRerank(info *relaycommon.RelayInfo, request dto.RerankRequest) {
+func (a *Adaptor) ConvertAudioRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.AudioRequest) (io.Reader, error) {
 	//TODO implement me
-
+	return nil, errors.New("not implemented")
 }
 
-func (a *Adaptor) Init(info *relaycommon.RelayInfo, request dto.GeneralOpenAIRequest) {
+func (a *Adaptor) ConvertImageRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.ImageRequest) (any, error) {
+	//TODO implement me
+	return nil, errors.New("not implemented")
+}
+
+func (a *Adaptor) Init(info *relaycommon.RelayInfo) {
 }
 
 func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {

relay/channel/perplexity/adaptor.go

@@ -15,12 +15,17 @@ import (
 type Adaptor struct {
 }
 
-func (a *Adaptor) InitRerank(info *relaycommon.RelayInfo, request dto.RerankRequest) {
+func (a *Adaptor) ConvertAudioRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.AudioRequest) (io.Reader, error) {
 	//TODO implement me
-
+	return nil, errors.New("not implemented")
 }
 
-func (a *Adaptor) Init(info *relaycommon.RelayInfo, request dto.GeneralOpenAIRequest) {
+func (a *Adaptor) ConvertImageRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.ImageRequest) (any, error) {
+	//TODO implement me
+	return nil, errors.New("not implemented")
+}
+
+func (a *Adaptor) Init(info *relaycommon.RelayInfo) {
 }
 
 func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {

relay/channel/tencent/adaptor.go

@@ -23,12 +23,17 @@ type Adaptor struct {
 	Timestamp int64
 }
 
-func (a *Adaptor) InitRerank(info *relaycommon.RelayInfo, request dto.RerankRequest) {
+func (a *Adaptor) ConvertAudioRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.AudioRequest) (io.Reader, error) {
 	//TODO implement me
-
+	return nil, errors.New("not implemented")
 }
 
-func (a *Adaptor) Init(info *relaycommon.RelayInfo, request dto.GeneralOpenAIRequest) {
+func (a *Adaptor) ConvertImageRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.ImageRequest) (any, error) {
+	//TODO implement me
+	return nil, errors.New("not implemented")
+}
+
+func (a *Adaptor) Init(info *relaycommon.RelayInfo) {
 	a.Action = "ChatCompletions"
 	a.Version = "2023-09-01"
 	a.Timestamp = common.GetTimestamp()

relay/channel/xunfei/adaptor.go

@@ -16,12 +16,17 @@ type Adaptor struct {
 	request *dto.GeneralOpenAIRequest
 }
 
-func (a *Adaptor) InitRerank(info *relaycommon.RelayInfo, request dto.RerankRequest) {
+func (a *Adaptor) ConvertAudioRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.AudioRequest) (io.Reader, error) {
 	//TODO implement me
-
+	return nil, errors.New("not implemented")
 }
 
-func (a *Adaptor) Init(info *relaycommon.RelayInfo, request dto.GeneralOpenAIRequest) {
+func (a *Adaptor) ConvertImageRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.ImageRequest) (any, error) {
+	//TODO implement me
+	return nil, errors.New("not implemented")
+}
+
+func (a *Adaptor) Init(info *relaycommon.RelayInfo) {
 }
 
 func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {

relay/channel/zhipu/adaptor.go

@@ -14,12 +14,17 @@ import (
 type Adaptor struct {
 }
 
-func (a *Adaptor) InitRerank(info *relaycommon.RelayInfo, request dto.RerankRequest) {
+func (a *Adaptor) ConvertAudioRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.AudioRequest) (io.Reader, error) {
 	//TODO implement me
-
+	return nil, errors.New("not implemented")
 }
 
-func (a *Adaptor) Init(info *relaycommon.RelayInfo, request dto.GeneralOpenAIRequest) {
+func (a *Adaptor) ConvertImageRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.ImageRequest) (any, error) {
+	//TODO implement me
+	return nil, errors.New("not implemented")
+}
+
+func (a *Adaptor) Init(info *relaycommon.RelayInfo) {
 }
 
 func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {

relay/channel/zhipu_4v/adaptor.go

@@ -15,12 +15,17 @@ import (
 type Adaptor struct {
 }
 
-func (a *Adaptor) InitRerank(info *relaycommon.RelayInfo, request dto.RerankRequest) {
+func (a *Adaptor) ConvertAudioRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.AudioRequest) (io.Reader, error) {
 	//TODO implement me
-
+	return nil, errors.New("not implemented")
 }
 
-func (a *Adaptor) Init(info *relaycommon.RelayInfo, request dto.GeneralOpenAIRequest) {
+func (a *Adaptor) ConvertImageRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.ImageRequest) (any, error) {
+	//TODO implement me
+	return nil, errors.New("not implemented")
+}
+
+func (a *Adaptor) Init(info *relaycommon.RelayInfo) {
 }
 
 func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {