openai tts

pkg/azureopenai/azureopenai.go

@@ -16,6 +16,9 @@ https://learn.microsoft.com/zh-cn/rest/api/cognitiveservices/azureopenaistable/m
   -H "Content-Type: application/json" \
   -H "api-key: $AZURE_OPENAI_KEY" \
 
+> GPT-4 Turbo
+https://techcommunity.microsoft.com/t5/ai-azure-ai-services-blog/azure-openai-service-launches-gpt-4-turbo-and-gpt-3-5-turbo-1106/ba-p/3985962
+
 */
 
 package azureopenai

pkg/openai/tts.go

@@ -0,0 +1,93 @@
+package openai
+
+import (
+	"bytes"
+	"encoding/json"
+	"fmt"
+	"io"
+	"log"
+	"net/http"
+	"net/http/httputil"
+	"net/url"
+	"opencatd-open/pkg/tokenizer"
+	"opencatd-open/store"
+
+	"github.com/gin-gonic/gin"
+)
+
+const (
+	SpeechEndpoint = "https://api.openai.com/v1/audio/speech"
+)
+
+type SpeechRequest struct {
+	Model string `json:"model"`
+	Input string `json:"input"`
+	Voice string `json:"voice"`
+}
+
+func SpeechHandler(c *gin.Context) {
+	var chatreq SpeechRequest
+	if err := c.ShouldBindJSON(&chatreq); err != nil {
+		c.JSON(http.StatusBadRequest, gin.H{"error": err.Error()})
+		return
+	}
+
+	var chatlog store.Tokens
+	chatlog.Model = chatreq.Model
+	chatlog.CompletionCount = len(chatreq.Input)
+
+	token, _ := c.Get("localuser")
+
+	lu, err := store.GetUserByToken(token.(string))
+	if err != nil {
+		c.JSON(http.StatusInternalServerError, gin.H{
+			"error": gin.H{
+				"message": err.Error(),
+			},
+		})
+		return
+	}
+	chatlog.UserID = int(lu.ID)
+
+	key, err := store.SelectKeyCache("openai")
+	if err != nil {
+		c.JSON(http.StatusInternalServerError, gin.H{
+			"error": gin.H{
+				"message": err.Error(),
+			},
+		})
+		return
+	}
+
+	targetURL, _ := url.Parse(SpeechEndpoint)
+	proxy := httputil.NewSingleHostReverseProxy(targetURL)
+
+	proxy.Director = func(req *http.Request) {
+		req.Header = c.Request.Header
+		req.Header["Authorization"] = []string{"Bearer " + key.Key}
+		req.Host = targetURL.Host
+		req.URL.Scheme = targetURL.Scheme
+		req.URL.Host = targetURL.Host
+		req.URL.Path = targetURL.Path
+		req.URL.RawPath = targetURL.RawPath
+
+		reqBytes, _ := json.Marshal(chatreq)
+		req.Body = io.NopCloser(bytes.NewReader(reqBytes))
+		req.ContentLength = int64(len(reqBytes))
+
+	}
+	proxy.ModifyResponse = func(resp *http.Response) error {
+		if resp.StatusCode == http.StatusOK {
+			chatlog.TotalTokens = chatlog.PromptCount + chatlog.CompletionCount
+			chatlog.Cost = fmt.Sprintf("%.6f", tokenizer.Cost(chatlog.Model, chatlog.PromptCount, chatlog.CompletionCount))
+			if err := store.Record(&chatlog); err != nil {
+				log.Println(err)
+			}
+			if err := store.SumDaily(chatlog.UserID); err != nil {
+				log.Println(err)
+			}
+		}
+		return nil
+	}
+	proxy.ServeHTTP(c.Writer, c.Request)
+}

pkg/tokenizer/tokenizer.go

@@ -107,7 +107,7 @@ func Cost(model string, promptCount, completionCount int) float64 {
 		cost = 11.02/1000000*float64(prompt) + (32.68/1000000)*float64(completion)
 	case "claude-instant-v1", "claude-instant-v1-100k":
 		cost = (1.63/1000000)*float64(prompt) + (5.51/1000000)*float64(completion)
-	case "claude-2":
+	case "claude-2", "claude-2.1":
 		cost = (11.02/1000000)*float64(prompt) + (32.68/1000000)*float64(completion)
 	default:
 		if strings.Contains(model, "gpt-3.5-turbo") {

router/router.go

@@ -17,6 +17,7 @@ import (
 	"net/url"
 	"opencatd-open/pkg/azureopenai"
 	"opencatd-open/pkg/claude"
+	oai "opencatd-open/pkg/openai"
 	"opencatd-open/pkg/tokenizer"
 	"opencatd-open/store"
 	"os"
@@ -480,6 +481,10 @@ func HandleProy(c *gin.Context) {
 		WhisperProxy(c)
 		return
 	}
+	if c.Request.URL.Path == "/v1/audio/speech" {
+		oai.SpeechHandler(c)
+		return
+	}
 
 	if c.Request.URL.Path == "/v1/chat/completions" && localuser {
 		if store.KeysCache.ItemCount() == 0 {