diff --git a/dto/audio.go b/dto/audio.go
index c56d6573d..ea51516f8 100644
--- a/dto/audio.go
+++ b/dto/audio.go
@@ -1,19 +1,22 @@
 package dto
 
 import (
+	"encoding/json"
+
 	"github.com/QuantumNous/new-api/types"
 
 	"github.com/gin-gonic/gin"
 )
 
 type AudioRequest struct {
-	Model          string  `json:"model"`
-	Input          string  `json:"input"`
-	Voice          string  `json:"voice"`
-	Instructions   string  `json:"instructions,omitempty"`
-	ResponseFormat string  `json:"response_format,omitempty"`
-	Speed          float64 `json:"speed,omitempty"`
-	StreamFormat   string  `json:"stream_format,omitempty"`
+	Model          string          `json:"model"`
+	Input          string          `json:"input"`
+	Voice          string          `json:"voice"`
+	Instructions   string          `json:"instructions,omitempty"`
+	ResponseFormat string          `json:"response_format,omitempty"`
+	Speed          float64         `json:"speed,omitempty"`
+	StreamFormat   string          `json:"stream_format,omitempty"`
+	Metadata       json.RawMessage `json:"metadata,omitempty"`
 }
 
 func (r *AudioRequest) GetTokenCountMeta() *types.TokenCountMeta {
diff --git a/relay/channel/volcengine/adaptor.go b/relay/channel/volcengine/adaptor.go
index 94cb776e8..411bdc8ef 100644
--- a/relay/channel/volcengine/adaptor.go
+++ b/relay/channel/volcengine/adaptor.go
@@ -75,6 +75,13 @@ func (a *Adaptor) ConvertAudioRequest(c *gin.Context, info *relaycommon.RelayInf
 		},
 	}
 
+	// 同步扩展字段的厂商自定义metadata
+	if len(request.Metadata) > 0 {
+		if err = json.Unmarshal(request.Metadata, &volcRequest); err != nil {
+			return nil, fmt.Errorf("error unmarshalling metadata to volcengine request: %w", err)
+		}
+	}
+
 	jsonData, err := json.Marshal(volcRequest)
 	if err != nil {
 		return nil, fmt.Errorf("error marshalling volcengine request: %w", err)