Right now, when using audio_input to interact with AI agent, we need to press the record button to start recording the audio. It will be great to have a keyword like VAD_threshold to enable voice activity detection above the threshold. When the threshold is 0, then the VAD is disabled and we still need to push down the button to start recording.
Related topics
| Topic | Replies | Views | Activity | |
|---|---|---|---|---|
| Streamlit Real Time audio Recorder with silence detection | 1 | 451 | October 21, 2025 | |
| Record sound from the user's microphone with streamlit | 9 | 40395 | June 9, 2023 | |
| New Component: streamlit-realtime-audio | 1 | 248 | January 14, 2026 | |
| Experience report: working with realtime audio in Streamlit | 1 | 696 | May 26, 2025 | |
| Speech to text in st.chat_input | 3 | 12699 | February 17, 2024 |