feat: doubao tts (labring#5285)

service/aiproxy/main.go

@@ -171,6 +171,7 @@ func main() {
 
 	<-ctx.Done()
 	log.Info("shutting down server...")
+	log.Info("max wait time: 120s")
 
 	shutdownCtx, cancel := context.WithTimeout(context.Background(), 120*time.Second)
 	defer cancel()

service/aiproxy/relay/adaptor/doubaoaudio/constants.go

@@ -0,0 +1,105 @@
+package doubaoaudio
+
+import (
+	"github.com/labring/sealos/service/aiproxy/model"
+	"github.com/labring/sealos/service/aiproxy/relay/relaymode"
+)
+
+// https://www.volcengine.com/docs/6561/1257543
+
+var ModelList = []*model.ModelConfig{
+	{
+		Model:      "Doubao-tts",
+		Type:       relaymode.AudioSpeech,
+		Owner:      model.ModelOwnerDoubao,
+		InputPrice: 0.5,
+		Config: map[model.ModelConfigKey]any{
+			model.ModelConfigSupportFormatsKey: []string{
+				"pcm",
+				"mp3",
+				"wav",
+				"ogg_opus",
+			},
+			model.ModelConfigSupportVoicesKey: []string{
+				"zh_female_cancan_mars_bigtts",
+				"zh_female_qingxinnvsheng_mars_bigtts",
+				"zh_female_shuangkuaisisi_moon_bigtts",
+				"zh_male_wennuanahu_moon_bigtts",
+				"zh_male_shaonianzixin_moon_bigtts",
+				"zh_female_zhixingnvsheng_mars_bigtts",
+				"zh_male_qingshuangnanda_mars_bigtts",
+				"zh_female_linjianvhai_moon_bigtts",
+				"zh_male_yuanboxiaoshu_moon_bigtts",
+				"zh_male_yangguangqingnian_moon_bigtts",
+				"zh_female_tianmeixiaoyuan_moon_bigtts",
+				"zh_female_qingchezizi_moon_bigtts",
+				"zh_male_jieshuoxiaoming_moon_bigtts",
+				"zh_female_kailangjiejie_moon_bigtts",
+				"zh_male_linjiananhai_moon_bigtts",
+				"zh_female_tianmeiyueyue_moon_bigtts",
+				"zh_female_xinlingjitang_moon_bigtts",
+				"en_male_smith_mars_bigtts",
+				"en_female_anna_mars_bigtts",
+				"en_male_adam_mars_bigtts",
+				"en_female_sarah_mars_bigtts",
+				"en_male_dryw_mars_bigtts",
+				"multi_male_jingqiangkanye_moon_bigtts",
+				"multi_female_shuangkuaisisi_moon_bigtts",
+				"multi_male_wanqudashu_moon_bigtts",
+				"multi_female_gaolengyujie_moon_bigtts",
+				"zh_male_jingqiangkanye_moon_bigtts",
+				"zh_female_wanwanxiaohe_moon_bigtts",
+				"zh_female_wanqudashu_moon_bigtts",
+				"zh_female_daimengchuanmei_moon_bigtts",
+				"zh_male_guozhoudege_moon_bigtts",
+				"zh_male_beijingxiaoye_moon_bigtts",
+				"zh_male_haoyuxiaoge_moon_bigtts",
+				"zh_male_guangxiyuanzhou_moon_bigtts",
+				"zh_female_meituojieer_moon_bigtts",
+				"zh_male_yuzhouzixuan_moon_bigtts",
+				"zh_male_naiqimengwa_mars_bigtts",
+				"zh_female_popo_mars_bigtts",
+				"zh_female_gaolengyujie_moon_bigtts",
+				"zh_male_aojiaobazong_moon_bigtts",
+				"zh_female_meilinvyou_moon_bigtts",
+				"zh_male_shenyeboke_moon_bigtts",
+				"zh_female_sajiaonvyou_moon_bigtts",
+				"zh_female_yuanqinvyou_moon_bigtts",
+				"ICL_zh_female_bingruoshaonv_tob",
+				"ICL_zh_female_huoponvhai_tob",
+				"zh_male_dongfanghaoran_moon_bigtts",
+				"ICL_zh_female_heainainai_tob",
+				"ICL_zh_female_linjuayi_tob",
+				"zh_female_wenrouxiaoya_moon_bigtts",
+				"zh_male_tiancaitongsheng_mars_bigtts",
+				"zh_male_sunwukong_mars_bigtts",
+				"zh_male_xionger_mars_bigtts",
+				"zh_female_peiqi_mars_bigtts",
+				"zh_female_wuzetian_mars_bigtts",
+				"zh_female_gujie_mars_bigtts",
+				"zh_female_yingtaowanzi_mars_bigtts",
+				"zh_male_chunhui_mars_bigtts",
+				"zh_female_shaoergushi_mars_bigtts",
+				"zh_male_silang_mars_bigtts",
+				"zh_male_jieshuonansheng_mars_bigtts",
+				"zh_female_jitangmeimei_mars_bigtts",
+				"zh_female_tiexinnvsheng_mars_bigtts",
+				"zh_female_qiaopinvsheng_mars_bigtts",
+				"zh_female_mengyatou_mars_bigtts",
+				"zh_male_changtianyi_mars_bigtts",
+				"zh_male_ruyaqingnian_mars_bigtts",
+				"zh_male_baqiqingshu_mars_bigtts",
+				"zh_male_qingcang_mars_bigtts",
+				"zh_female_gufengshaoyu_mars_bigtts",
+				"zh_female_wenroushunv_mars_bigtts",
+			},
+		},
+	},
+
+	// {
+	// 	Model:      "Doubao-stt",
+	// 	Type:       relaymode.AudioTranscription,
+	// 	Owner:      model.ModelOwnerDoubao,
+	// 	InputPrice: 2.3,
+	// },
+}

service/aiproxy/relay/adaptor/doubaoaudio/main.go

@@ -0,0 +1,99 @@
+package doubaoaudio
+
+import (
+	"errors"
+	"fmt"
+	"io"
+	"net/http"
+	"strings"
+
+	"github.com/gin-gonic/gin"
+	"github.com/labring/sealos/service/aiproxy/model"
+	"github.com/labring/sealos/service/aiproxy/relay/adaptor/openai"
+	"github.com/labring/sealos/service/aiproxy/relay/meta"
+	relaymodel "github.com/labring/sealos/service/aiproxy/relay/model"
+	"github.com/labring/sealos/service/aiproxy/relay/relaymode"
+)
+
+func GetRequestURL(meta *meta.Meta) (string, error) {
+	u := meta.Channel.BaseURL
+	if u == "" {
+		u = baseURL
+	}
+	switch meta.Mode {
+	case relaymode.AudioSpeech:
+		return u + "/api/v1/tts/ws_binary", nil
+	default:
+		return "", fmt.Errorf("unsupported relay mode %d for doubao", meta.Mode)
+	}
+}
+
+type Adaptor struct{}
+
+const baseURL = "https://openspeech.bytedance.com"
+
+func (a *Adaptor) GetModelList() []*model.ModelConfig {
+	return ModelList
+}
+
+func (a *Adaptor) GetRequestURL(meta *meta.Meta) (string, error) {
+	return GetRequestURL(meta)
+}
+
+func (a *Adaptor) ConvertRequest(meta *meta.Meta, req *http.Request) (http.Header, io.Reader, error) {
+	switch meta.Mode {
+	case relaymode.AudioSpeech:
+		return ConvertTTSRequest(meta, req)
+	default:
+		return nil, nil, fmt.Errorf("unsupported relay mode %d for doubao", meta.Mode)
+	}
+}
+
+// key格式: app_id|app_token
+func getAppIDAndToken(key string) (string, string, error) {
+	parts := strings.Split(key, "|")
+	if len(parts) != 2 {
+		return "", "", errors.New("invalid key format")
+	}
+	return parts[0], parts[1], nil
+}
+
+func (a *Adaptor) SetupRequestHeader(meta *meta.Meta, _ *gin.Context, req *http.Request) error {
+	switch meta.Mode {
+	case relaymode.AudioSpeech:
+		_, token, err := getAppIDAndToken(meta.Channel.Key)
+		if err != nil {
+			return err
+		}
+		req.Header.Set("Authorization", "Bearer;"+token)
+		return nil
+	default:
+		return fmt.Errorf("unsupported relay mode %d for doubao", meta.Mode)
+	}
+}
+
+func (a *Adaptor) DoRequest(meta *meta.Meta, _ *gin.Context, req *http.Request) (*http.Response, error) {
+	switch meta.Mode {
+	case relaymode.AudioSpeech:
+		return TTSDoRequest(meta, req)
+	default:
+		return nil, fmt.Errorf("unsupported relay mode %d for doubao", meta.Mode)
+	}
+}
+
+func (a *Adaptor) DoResponse(meta *meta.Meta, c *gin.Context, resp *http.Response) (*relaymodel.Usage, *relaymodel.ErrorWithStatusCode) {
+	switch meta.Mode {
+	case relaymode.AudioSpeech:
+		return TTSDoResponse(meta, c, resp)
+	default:
+		return nil, openai.ErrorWrapperWithMessage(
+			fmt.Sprintf("unsupported relay mode %d for doubao", meta.Mode),
+			nil,
+			http.StatusBadRequest,
+		)
+	}
+}
+
+func (a *Adaptor) GetChannelName() string {
+	return "doubao audio"
+}