Problem: The speech recognition of Whisper-1 does not work on the iPhone and iPad.

[andreaswittmann]

[2025-03-31 Mon 12:37] I have succesfully configured the Whisper-1 API from OpenAI in the LibreChat configuration file.
The STT works on some devices but mostly fails on Apple technology.

Here is table with my tests:

Device	Browser	Result
Macbook Pro	Safari	Fails
Macbook Pro	Chrome	Works
iPhone 14 Pro	Safari	Fails
iPhone 14 Pro	Firefox	Fails
iPhone 14 Pro	Chrome	Fails
iPad Pro	Safari	Fails
iPad Pro	Chrome	Fails
Microsoft Surface Pro 8	Edge	Works
Microsoft Surface Pro 8	Chrome	Works

I have analyszed the error log.

The error from the error.log is to big to paste it in here. But here is the analysis of Antrophic-Claude:

# Analysis of the OpenAI API Error

## Root Cause
The error is related to an invalid file format when trying to transcribe audio using OpenAI's Speech-to-Text API. 

The specific error message in the response is:
```json
"message": "Invalid file format. Supported formats: ['flac', 'm4a', 'mp3', 'mp4', 'mpeg', 'mpga', 'oga', 'ogg', 'wav', 'webm']"
```

## What's Happening
1. Your code is attempting to make a POST request to OpenAI's audio transcription endpoint: `https://api.openai.com/v1/audio/transcriptions`
2. You're sending a file named `audio.m4a` in the request
3. Despite the `.m4a` extension (which is a supported format according to the error message), the API is rejecting the file because it's not recognizing it as a valid audio format

## Possible Issues
1. **File Content Mismatch**: The file might have an `.m4a` extension but could be corrupted or contain data in a different format
2. **File Reading Problem**: The file may not be getting properly read or uploaded in the request
3. **Incomplete File**: The request data shows `"_valueLength": 9` which seems unusually small for an audio file, suggesting the file content might not be fully loaded
4. **Content Type Issues**: The file may need a specific MIME type to be properly recognized

My suspicion is that it is a problem with the audio codec used by Apple devices.

Can anyone confirm this behavior?
Do you have any suggestions for a temporary workaround?
Or do you have any other ideas how to solve this problem?