PyInstaller executable throws “OSError: [WinError 50] The request is not supported” when using speech_recognition (FLAC error)

I’m building a Python voice assistant using the speech_recognition library.
Everything works perfectly when I run the code from PyCharm or the terminal,
but when I convert it to an .exe using Auto Py to Exe / PyInstaller,
I get this error as soon as the recognizer tries to process audio:

pygame 2.6.1 (SDL 2.28.4, Python 3.12.1) 
Welcome to Vision! 
Say any of the models name to activate. 
Say 'exit' to quit. 
Waiting for wake word... Listening... Exception in thread Thread-1 (chatbot): 
Traceback (most recent call last):   
File "threading.py", line 1073, in _bootstrap_inner   File "threading.py", line 1010, in run   
File "vision_v3.py", line 126, in chatbot   
File "jarvis_functions\essential_functions\voice_input.py", line 47, in record_text   
File "speech_recognition\recognizers\google.py", line 253, in recognize_legacy   
File "speech_recognition\recognizers\google.py", line 56, in build   
File "speech_recognition\recognizers\google.py", line 88, in build_data   
File "speech_recognition\audio.py", line 256, in get_flac_data   
File "subprocess.py", line 992, in __init__   
File "subprocess.py", line 1407, in _get_handles   
File "subprocess.py", line 1416, in _make_inheritable 
OSError: [WinError 50] The request is not supported 

I've tried to include all my folders and .env file with Auto Py to Exe:

--add-data "D:\pycharm\opencv\Vision\jarvis_functions;jarvis_functions" --add-data "D:\pycharm\opencv\Vision\sound_files\beep;sound_files/beep" --add-data "D:\pycharm\opencv\Vision\sound_files\cam;sound_files/cam"

Converting audio to WAV manually using:

wav_data = audio.get_wav_data()

Verified that everything works in Python directly, but the .exe fails.

This is a simplified code snippet:

import speech_recognition as sr  
r = sr.Recognizer()   
def record_text(): 
    try:         
        with sr.Microphone() as source:             
        r.adjust_for_ambient_noise(source, duration=0.2)             
        audio = r.listen(source)             
        text = r.recognize_google(audio, language="bg-BG")             
        return text.lower()     
    except sr.RequestError as e:         
        print(f"API error: {e}")     
    except sr.UnknownValueError:         
        print("Could not understand audio.") 

Why does the PyInstaller .exe version fail with [WinError 50] inside get_flac_data() when speech_recognition works perfectly in normal Python?

Do I need to bundle a flac.exe file manually, or is there a better way to make speech_recognition use WAV instead of FLAC so it works in compiled executables?