feat: 支持 MiniMax 音色克隆和 2.8 语音模型

2026-06-17 02:13:21 +08:00
parent 02ba5d3cdd
commit c4341335d7
20 changed files with 1645 additions and 10 deletions
@@ -182,6 +182,88 @@ func TestMinimaxClientSpeechUsesT2AV2AndNormalizesAudio(t *testing.T) {
 	}
 }

+func TestMinimaxVoiceCloneTextValidationPayload(t *testing.T) {
+	var capturedClone map[string]any
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		if got := r.Header.Get("Authorization"); got != "Bearer test-key" {
+			t.Fatalf("unexpected auth header: %q", got)
+		}
+		w.Header().Set("Content-Type", "application/json")
+		switch r.URL.Path {
+		case "/files/upload":
+			_ = json.NewEncoder(w).Encode(map[string]any{
+				"file":      map[string]any{"file_id": "123456"},
+				"base_resp": map[string]any{"status_code": 0},
+			})
+		case "/voice_clone":
+			if err := json.NewDecoder(r.Body).Decode(&capturedClone); err != nil {
+				t.Fatalf("decode voice clone request: %v", err)
+			}
+			_ = json.NewEncoder(w).Encode(map[string]any{
+				"demo_audio": "",
+				"base_resp":  map[string]any{"status_code": 0},
+			})
+		default:
+			t.Fatalf("unexpected request: %s %s", r.Method, r.URL.String())
+		}
+	}))
+	defer server.Close()
+
+	_, err := (MinimaxClient{HTTPClient: server.Client()}).Run(context.Background(), Request{
+		Kind:  "voice.clone",
+		Model: "MiniMax-Voice-Clone",
+		Body: map[string]any{
+			"voice_id":                  "voice_test_123",
+			"audio_url":                 "data:audio/wav;base64," + base64.StdEncoding.EncodeToString([]byte("wave")),
+			"text_validation":           false,
+			"need_noise_reduction":      true,
+			"need_volume_normalization": true,
+			"aigc_watermark":            false,
+		},
+		Candidate: store.RuntimeModelCandidate{
+			Provider:          "minimax",
+			BaseURL:           server.URL,
+			ProviderModelName: "voice_clone",
+			Credentials:       map[string]any{"apiKey": "test-key"},
+		},
+	})
+	if err != nil {
+		t.Fatalf("run minimax voice clone client: %v", err)
+	}
+	if _, ok := capturedClone["text_validation"]; ok {
+		t.Fatalf("legacy boolean text_validation should be omitted: %+v", capturedClone)
+	}
+	if capturedClone["file_id"] != float64(123456) {
+		t.Fatalf("file_id should be submitted as number: %+v", capturedClone)
+	}
+
+	capturedClone = nil
+	_, err = (MinimaxClient{HTTPClient: server.Client()}).Run(context.Background(), Request{
+		Kind:  "voice.clone",
+		Model: "MiniMax-Voice-Clone",
+		Body: map[string]any{
+			"voice_id":        "voice_test_456",
+			"audio_url":       "data:audio/wav;base64," + base64.StdEncoding.EncodeToString([]byte("wave")),
+			"text_validation": "  这是一段用于校验的源音频文本  ",
+		},
+		Candidate: store.RuntimeModelCandidate{
+			Provider:          "minimax",
+			BaseURL:           server.URL,
+			ProviderModelName: "voice_clone",
+			Credentials:       map[string]any{"apiKey": "test-key"},
+		},
+	})
+	if err != nil {
+		t.Fatalf("run minimax voice clone client with transcript: %v", err)
+	}
+	if capturedClone["text_validation"] != "这是一段用于校验的源音频文本" {
+		t.Fatalf("unexpected text_validation payload: %+v", capturedClone)
+	}
+	if capturedClone["file_id"] != float64(123456) {
+		t.Fatalf("file_id should be submitted as number with transcript: %+v", capturedClone)
+	}
+}
+
 func TestSimulationDurationCanBeControlledByParams(t *testing.T) {
 	fixedDuration := simulationDuration(Request{Body: map[string]any{"simulationDurationSeconds": 7}})
 	if fixedDuration != 7*time.Second {