diff --git a/metagpt/learn/text_to_embedding.py b/metagpt/learn/text_to_embedding.py
index b1395a61a..281815ca6 100644
--- a/metagpt/learn/text_to_embedding.py
+++ b/metagpt/learn/text_to_embedding.py
@@ -6,8 +6,9 @@
 @File    : text_to_embedding.py
 @Desc    : Text-to-Embedding skill, which provides text-to-embedding functionality.
 """
+import os
 
-from metagpt.tools.openai_text_2_embedding import oas3_openai_text_2_embedding
+from metagpt.tools.openai_text_to_embedding import oas3_openai_text_to_embedding
 from metagpt.utils.common import initialize_environment
 
 
@@ -20,4 +21,6 @@ def text_to_embedding(text, model="text-embedding-ada-002", openai_api_key=""):
     :return: A json object of :class:`ResultEmbedding` class if successful, otherwise `{}`.
     """
     initialize_environment()
-    return oas3_openai_text_2_embedding(text, model=model, openai_api_key=openai_api_key)
\ No newline at end of file
+    if os.environ.get("OPENAI_API_KEY") or openai_api_key:
+        return oas3_openai_text_to_embedding(text, model=model, openai_api_key=openai_api_key)
+    raise EnvironmentError
diff --git a/metagpt/learn/text_to_image.py b/metagpt/learn/text_to_image.py
index 87668a13f..0932dfe07 100644
--- a/metagpt/learn/text_to_image.py
+++ b/metagpt/learn/text_to_image.py
@@ -6,18 +6,25 @@
 @File    : text_to_image.py
 @Desc    : Text-to-Image skill, which provides text-to-image functionality.
 """
+import os
 
-from metagpt.tools.openai_text_2_image import oas3_openai_text_2_image
+from metagpt.tools.metagpt_text_to_image import oas3_metagpt_text_to_image
+from metagpt.tools.openai_text_to_image import oas3_openai_text_to_image
 from metagpt.utils.common import initialize_environment
 
 
-def text_to_image(text, size_type: str = "1024x1024", openai_api_key=""):
+def text_to_image(text, size_type: str = "512x512", openai_api_key="", model_url=""):
     """Text to image
 
     :param text: The text used for image conversion.
     :param openai_api_key: OpenAI API key, For more details, checkout: `https://platform.openai.com/account/api-keys`
-    :param size_type: One of ['256x256', '512x512', '1024x1024']
+    :param size_type: If using OPENAI, the available size options are ['256x256', '512x512', '1024x1024'], while for MetaGPT, the options are ['512x512', '512x768'].
+    :param model_url: MetaGPT model url
     :return: The image data is returned in Base64 encoding.
     """
     initialize_environment()
-    return oas3_openai_text_2_image(text, size_type, openai_api_key)
+    if os.environ.get("METAGPT_TEXT_TO_IMAGE_MODEL") or model_url:
+        return oas3_metagpt_text_to_image(text, size_type, model_url)
+    if os.environ.get("OPENAI_API_KEY") or openai_api_key:
+        return oas3_openai_text_to_image(text, size_type, openai_api_key)
+    raise EnvironmentError
diff --git a/metagpt/learn/text_to_speech.py b/metagpt/learn/text_to_speech.py
index 909a9dca1..b89b5a9c4 100644
--- a/metagpt/learn/text_to_speech.py
+++ b/metagpt/learn/text_to_speech.py
@@ -6,6 +6,7 @@
 @File    : text_to_speech.py
 @Desc    : Text-to-Speech skill, which provides text-to-speech functionality
 """
+import os
 
 from metagpt.tools.azure_tts import oas3_azsure_tts
 from metagpt.utils.common import initialize_environment
@@ -26,4 +27,8 @@ def text_to_speech(text, lang="zh-CN", voice="zh-CN-XiaomoNeural", style="affect
 
     """
     initialize_environment()
-    return oas3_azsure_tts(text, lang, voice, style, role, subscription_key, region)
+    if (os.environ.get("AZURE_TTS_SUBSCRIPTION_KEY") and os.environ.get("AZURE_TTS_REGION")) or \
+        (subscription_key and region):
+        return oas3_azsure_tts(text, lang, voice, style, role, subscription_key, region)
+
+    raise EnvironmentError
diff --git a/tests/metagpt/learn/test_text_to_image.py b/tests/metagpt/learn/test_text_to_image.py
index bfcb1db25..545c8a3ef 100644
--- a/tests/metagpt/learn/test_text_to_image.py
+++ b/tests/metagpt/learn/test_text_to_image.py
@@ -20,7 +20,7 @@ async def mock_text_to_image():
         size_type: str
 
     inputs = [
-        {"input": "Panda emoji", "size_type": "256x256"}
+        {"input": "Panda emoji", "size_type": "512x512"}
     ]
 
     for i in inputs: