Outdated event listening in `openai_stt.py`

[majklost]

Bug description

when running example examples/voice/streamed, the transcription is never processed, because it expects an event type input_audio_transcription_completed. After update of the API, the correct event to listen for is conversation.item.input_audio_transcription.completed

Debug information

• Agents SDK version: (e.g. v0.2.11)
• Python version (e.g. Python 3.12.2)

Repro steps

running

python -m examples.voice.streamed.main

no output will be generated. After the suggested fix, example runs normally.

[seratch]

Thanks for reporting this issue. #1537 will resolve this issue.