Server can not process request when sending audio realtime api

Question

Server can not process request when sending audio realtime api

Alta Saunders 25

When sending audio bytes with the request (input_audio_buffer.append) for the realtime api, I get the following error:

{'type': 'error', 'event_id': 'event_BbLkzxTgSJvg6hrELg7qk', 'error': {'type': 'server_error', 'code': None, 'message': 'The server had an error while processing your request. Sorry about that! Please contact us by filing an Azure support request in the portal. You can find information on how to do that here: https://learn.microsoft.com/azure/ai-services/cognitive-services-support-options if the error persists. (include session ID in your message: sess_BbLkzvkPQRZQhOmdHIjxk). We recommend you retry your request. If the problem persists, contact us by filing an Azure support request in the portal. You can find information on how to do that here: https://learn.microsoft.com/azure/ai-services/cognitive-services-support-options . (Please include the session ID sess_BbLkzvkPQRZQhOmdHIjxk in your message.)', 'param': None, 'event_id': None}}

However if I call the api through openai directly and not through the api for the model deployed on ai foundry, then I do not have any problems.

What is causing this problem?

Manas Mohanty 17,015 Reputation points Microsoft External Staff Moderator

2025-05-26T08:27:01.01+00:00

Hi Alta Saunders

Sorry for the trouble

Ask

Could you please share any sample SDK code along the documentation referred (mention model and region too) to replicate the issue at our side.

Could you also try in Azure AI foundry UI instead of SDK.

Thank you

Alta Saunders 25

Model being used: gpt-4o-mini-realtime-preview
region: Sweden Central

websockets api documentation: https://learn.microsoft.com/en-us/azure/ai-services/openai/realtime-audio-reference

sample code where call is made:

async def send_audio(self, audio_data):
        """Send audio input."""
        try:
            base64_chunk = base64.b64encode(audio_data).decode("utf-8")
            await self.send_event(
                {
                    "type": "input_audio_buffer.append",
                    "audio": base64_chunk,
                }
            )

        except Exception as e:
            threads4life_logger.error(f"Audio sending error: {e}")

Manas Mohanty 17,015 Reputation points Microsoft External Staff Moderator

2025-05-27T08:16:46.0333333+00:00

Hi Alta Saunders

Glad to find you have been able to find the parameter causing this issue.

I have provided info for valid values that can be provided to that param.

Reference - https://learn.microsoft.com/en-us/azure/ai-services/openai/realtime-audio-reference#realtimerequestsession

Thank you.
Alta Saunders 25 Reputation points

2025-05-27T09:07:44.9466667+00:00

Thank you. There seems to be differences in the api form openai and azure. Is there an update as to when they will be the same?
Manas Mohanty 17,015 Reputation points Microsoft External Staff Moderator

2025-05-27T09:26:28.0766667+00:00

Hi Alta Saunders

Yes, Azure OpenAI Syntax will be slightly different from OpenAI syntax due to integration and hosting in Azure Cloud, underlying backend changes from product team for same.

I don't think they will be 100% exact in near future.

But You can post the difference in usage as feedback here in feedback forum.

Please let us know if we can consider this case as resolved with above findings.

Thank you.
Alta Saunders 25 Reputation points

2025-05-27T09:35:46.0333333+00:00

Ok thank you. This cased can be considered as resolved

Answer accepted by question author

0 additional answers

Your answer

Manas Mohanty 17,015 Reputation points Microsoft External Staff Moderator

2025-05-26T08:27:01.01+00:00

Hi Alta Saunders

Sorry for the trouble

Ask

Could you please share any sample SDK code along the documentation referred (mention model and region too) to replicate the issue at our side.

Could you also try in Azure AI foundry UI instead of SDK.

Thank you
Alta Saunders 25 Reputation points

2025-05-26T08:32:03.6733333+00:00

Model being used: gpt-4o-mini-realtime-preview
region: Sweden Central

websockets api documentation: https://learn.microsoft.com/en-us/azure/ai-services/openai/realtime-audio-reference

sample code where call is made:

async def send_audio(self, audio_data): """Send audio input.""" try: base64_chunk = base64.b64encode(audio_data).decode("utf-8") await self.send_event( { "type": "input_audio_buffer.append", "audio": base64_chunk, } ) except Exception as e: threads4life_logger.error(f"Audio sending error: {e}")
Manas Mohanty 17,015 Reputation points Microsoft External Staff Moderator

2025-05-27T08:16:46.0333333+00:00

Hi Alta Saunders

Glad to find you have been able to find the parameter causing this issue.

I have provided info for valid values that can be provided to that param.

Reference - https://learn.microsoft.com/en-us/azure/ai-services/openai/realtime-audio-reference#realtimerequestsession

Thank you.
Alta Saunders 25 Reputation points

2025-05-27T09:07:44.9466667+00:00

Thank you. There seems to be differences in the api form openai and azure. Is there an update as to when they will be the same?
Manas Mohanty 17,015 Reputation points Microsoft External Staff Moderator

2025-05-27T09:26:28.0766667+00:00

Hi Alta Saunders

Yes, Azure OpenAI Syntax will be slightly different from OpenAI syntax due to integration and hosting in Azure Cloud, underlying backend changes from product team for same.

I don't think they will be 100% exact in near future.

But You can post the difference in usage as feedback here in feedback forum.

Please let us know if we can consider this case as resolved with above findings.

Thank you.
Alta Saunders 25 Reputation points

2025-05-27T09:35:46.0333333+00:00

Ok thank you. This cased can be considered as resolved

Answer 1

Hi Alta Saunders

Attached case summary for reference.

Issue

'error': {'type': 'server_error', 'code': None, 'message': 'The server had an error while processing your request. Sorry about that! Please contact us by filing an Azure support request in the portal.

while sending audio_chunks in real time api

Solution

Customer found the param responsible triggering server error

input_audio_noise_reduction"

Customer was suggested to use boolean value instead of dictionary value mentioned in Azure OpenAI.

Status: Customer is able to get results by modifying the concerned param.

Please take a moment to accept this answer if you acknowledged our inputs.

Thank you.

Share via

Server can not process request when sending audio realtime api

0 additional answers

Your answer