Realtime Recognizer not utilising with Semantic Segmentation

Hi all!

I'm using the Azure speechsdk.SpeechRecognizer for transcribing streamed real-time audio. While the transcription works, continuous talking will result in large paragraphs being outputted rather than sentence by sentence. I included the speech_config.set_property(speechsdk.PropertyId.Speech_SegmentationStrategy, "Semantic") in order to utilise the semantic separation rather than audio based separation. However the output is 100% identical. For example it took this long for the recognizer to output:
a classic example of early remote control the idea was simple club and your lights will turn on off you didn't need to physically handle a wire remote we didn't have to find a clicker to point at a TV and you could make automations happen at a very limited scale i remember that i'd clap along to a song and the lights would go wild

The output:

A classic example of early remote control. The idea was simple, club and your lights will turn on off. You didn't need to physically handle a wire remote, we didn't have to find a clicker to point at a TV, and you could make automations happen at a very limited scale. I remember that I'd clap along to a song and the lights would go wild.
The output is correct and I can see that the recognizer is collecting word by word but it took this whole paragraph until it was returned formatted. The behavior is exactly as if the semantic recognition is not used.

This is how I set up the recognizer, is there some step missing?

speech_config = speechsdk.SpeechConfig(subscription=os.getenv('AZURE_SPEECH_KEY'), region=os.getenv('AZURE_SPEECH_REGION'))
            speech_config.set_property(speechsdk.PropertyId.Speech_SegmentationStrategy, "Semantic")
            speech_config.speech_recognition_language="en-US"
            speech_config.set_property(speechsdk.PropertyId.Spee)
            self.push_stream = speechsdk.audio.PushAudioInputStream()
            audio_config = speechsdk.audio.AudioConfig(stream=self.push_stream)
            self.speech_recognizer = speechsdk.SpeechRecognizer(speech_config=speech_config, audio_config=audio_config)