microsoft/semantic-kernel · error · ServiceInitializationError

Failed to create OpenAI settings.

Error message

Failed to create OpenAI settings.

What it means

Raised by OpenAIAudioToText.__init__ when OpenAISettings construction throws a pydantic ValidationError. The non-Azure OpenAISettings model has fewer fields than its Azure counterpart: api_key (SecretStr|None), org_id (str|None), audio_to_text_model_id (str|None). If a field value fails type coercion (e.g. api_key passed as bytes or org_id as a non-string), pydantic raises ValidationError which is wrapped here.

Source

Thrown at python/semantic_kernel/connectors/ai/open_ai/services/open_ai_audio_to_text.py:58

            org_id: The optional org ID to use. If provided will override,
                the env vars or .env file value.
            default_headers: The default headers mapping of string keys to
                string values for HTTP requests. (Optional)
            async_client: An existing client to use. (Optional)
            env_file_path: Use the environment settings file as
                a fallback to environment variables. (Optional)
            env_file_encoding: The encoding of the environment settings file. (Optional)
        """
        try:
            openai_settings = OpenAISettings(
                api_key=api_key,
                org_id=org_id,
                audio_to_text_model_id=ai_model_id,
                env_file_path=env_file_path,
                env_file_encoding=env_file_encoding,
            )
        except ValidationError as ex:
            raise ServiceInitializationError("Failed to create OpenAI settings.", ex) from ex
        if not openai_settings.audio_to_text_model_id:
            raise ServiceInitializationError("The OpenAI audio to text model ID is required.")
        super().__init__(
            ai_model_id=openai_settings.audio_to_text_model_id,
            api_key=openai_settings.api_key.get_secret_value() if openai_settings.api_key else None,
            ai_model_type=OpenAIModelTypes.AUDIO_TO_TEXT,
            org_id=openai_settings.org_id,
            service_id=service_id,
            default_headers=default_headers,
            client=async_client,
        )

    @classmethod
    def from_dict(cls: type[T_], settings: dict[str, Any]) -> T_:
        """Initialize an Open AI service from a dictionary of settings.

        Args:
            settings: A dictionary of settings for the service.

View on GitHub (pinned to c028a0c7dc)

Solutions

1. Inspect the chained ValidationError for the specific field and constraint that failed.
2. Pass api_key as a plain string (str), not a SecretStr or bytes.
3. Verify your .env file syntax and encoding (default is utf-8).
4. Ensure OPENAI_API_KEY is set to a non-empty string in the environment.

Example fix

# before (passing SecretStr instead of str)
from pydantic import SecretStr
service = OpenAIAudioToText(
    ai_model_id='whisper-1',
    api_key=SecretStr('sk-...'),  # wrong type
)
# after (plain string)
service = OpenAIAudioToText(
    ai_model_id='whisper-1',
    api_key='sk-...',
)

import os

api_key = os.environ.get('OPENAI_API_KEY')
if not api_key or not isinstance(api_key, str):
    raise ValueError(
        'OPENAI_API_KEY must be set to a non-empty string in the environment or .env file.'
    )

from semantic_kernel.exceptions.service_exceptions import ServiceInitializationError

try:
    service = OpenAIAudioToText(
        ai_model_id='whisper-1',
        api_key=os.environ.get('OPENAI_API_KEY'),
    )
except ServiceInitializationError as e:
    cause = e.__cause__
    if cause:
        print(f'Settings validation failed: {cause}')
    raise

When it happens

Trigger: Constructing OpenAIAudioToText with a malformed .env file, an api_key of the wrong type (e.g. bytes instead of str), or an environment variable with an incompatible value type. Since most fields are str|None, type errors are rare but can occur with corrupt env files.

Common situations: Corrupted .env file with invalid syntax; OPENAI_API_KEY env var set to an empty string in CI; env_file_encoding mismatch causing decode errors; passing api_key as a SecretStr object instead of a plain string.

Related errors

• Failed to create OpenAI settings.
• The input audio content is not readable.
• The audio transcription format '{responseFormat}' is not sup
• The provided reasoning effort '{textEffortLevel}' is not sup
• When used with number_of_responses, best_of controls the num

AI-assisted analysis of microsoft/semantic-kernel@c028a0c7dc (2026-08-13). Data as JSON: /api/errors/e9e7b0a7d25c60a0. Report an issue: GitHub.