I’m using Solara to build a web app with Python, and I can use ipywebrtc to capture audio from the client’s browser. I can first record the audio to a temporary file and then pass it to Azure Speech, but I need it to be streaming

At first, I tried with this code

from ipywebrtc import CameraStream, AudioRecorder
from azure.cognitiveservices.speech.audio import AudioInputStream
import azure.cognitiveservices.speech as speechsdk
...
speech_config = speechsdk.SpeechConfig(subscription="...", region="brazilsouth")

camera = CameraStream(constraints={'audio': True, 'video': False})
recorder = AudioRecorder(stream=camera)
stream = AudioInputStream(recorder.audio)
audio_config = speechsdk.audio.AudioConfig(stream=stream)
conversation_transcriber = speechsdk.transcription.ConversationTranscriber(speech_config=speech_config,audio_config=audio_config)
...

Exception in thread Thread-22 (recognize_from_device):
Traceback (most recent call last):
  File "C:Python312Libthreading.py", line 1073, in _bootstrap_inner
    self.run()
  File "C:Python312Libthreading.py", line 1010, in run
    self._target(*self._args, **self._kwargs)
  File "L:projectstestesfunctionsouvir_microfone.py", line 62, in recognize_from_device
    audio_config = speechsdk.audio.AudioConfig(stream=stream)
                   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "L:projectstestesvenvLibsite-packagesazurecognitiveservicesspeechaudio.py", line 382, in __init__
    _call_hr_fn(fn=_sdk_lib.audio_config_create_audio_input_from_stream, *[ctypes.byref(handle), stream._handle])
  File "L:projectstestesvenvLibsite-packagesazurecognitiveservicesspeechinterop.py", line 61, in _call_hr_fn
    hr = fn(*args) if len(args) > 0 else fn()
         ^^^^^^^^^
ctypes.ArgumentError: argument 2: TypeError: Don't know how to convert parameter 2
Exception ignored in: <function _Handle.__del__ at 0x00000140E12B5E40>
Traceback (most recent call last):
  File "L:projectstestesvenvLibsite-packagesazurecognitiveservicesspeechinterop.py", line 105, in __del__
    elif self.__test_fn(self.__handle):
         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
ctypes.ArgumentError: argument 1: TypeError: Don't know how to convert parameter 1

changing AudioInputStream(recorder.audio) to AudioInputStream(recorder.codecs):

RuntimeError: Exception with error code:
[CALL STACK BEGIN]

    > GetModuleObject
    - audio_config_get_audio_processing_options
    - audio_config_create_audio_input_from_stream
    - ffi_prep_go_closure
    - ffi_call_go
    - ffi_call
    - 00007FF910133DD5 (SymFromAddr() error: Attempt to access invalid address.)
    - 00007FF910132D33 (SymFromAddr() error: Attempt to access invalid address.)
    - 00007FF910132928 (SymFromAddr() error: Attempt to access invalid address.)
    - PyObject_Call
    - PyEval_EvalFrameDefault
    - PyFunction_Vectorcall
    - PyObject_VectorcallMethod
    - PyObject_Vectorcall
    - PyObject_Vectorcall
    - PyEval_EvalFrameDefault

[CALL STACK END]

Exception with an error code: 0x5 (SPXERR_INVALID_ARG)