Please address the comments from this code review:

## Overall Comments
- The regex `^[a-zA-Z0-9]{32,84}$` allows any length between 32 and 84 characters, not exactly 32 or 84 as described; if you really only want 32 or 84, consider two alternatives like `^(?:[a-zA-Z0-9]{32}|[a-zA-Z0-9]{84})$`.

## Individual Comments

### Comment 1
<location path="astrbot/core/provider/sources/azure_tts_source.py" line_range="23-26" />
<code_context>

 TEMP_DIR = Path(get_astrbot_temp_path()) / "azure_tts"
 TEMP_DIR.mkdir(parents=True, exist_ok=True)
+AZURE_TTS_SUBSCRIPTION_KEY_PATTERN = r"^[a-zA-Z0-9]{32,84}$"


</code_context>
<issue_to_address>
**issue (bug_risk):** The regex `{32,84}` matches any length between 32 and 84, not "32 or 84" specifically as suggested by the error message.

This pattern allows any alphanumeric string 32–84 chars long, but the later message "32位或84位字母数字" implies only 32 or 84 chars are valid. If that’s the intent, use something like `r"^(?:[a-zA-Z0-9]{32}|[a-zA-Z0-9]{84})$"` instead of `{32,84}`.
</issue_to_address>

### Comment 2
<location path="tests/test_azure_tts_provider.py" line_range="6" />
<code_context>
+from astrbot.core.provider.sources.azure_tts_source import AzureTTSProvider
+
+
+@pytest.mark.parametrize("key_length", [32, 84])
+def test_azure_tts_accepts_subscription_keys_with_supported_lengths(key_length: int):
+    config = {
</code_context>
<issue_to_address>
**suggestion (testing):** Add negative tests for unsupported key lengths and non-alphanumeric keys

This only verifies the accepted lengths; it doesn’t protect against invalid keys accidentally becoming allowed. Please add a parametrized test that checks `AzureTTSProvider` raises `ValueError` for invalid lengths (e.g. 0, 31, 33, 83, 85) and for keys with non‑alphanumeric characters (e.g. `'-'`, `'_'`, `'~'`) to lock in both length and character-set validation.
</issue_to_address>