Problem
When TTS is enabled, the typing indicator (chat action) stops as soon as the model run completes, but the TTS audio generation and sending happens after in the dispatch phase.
On Telegram (and likely other platforms), this causes the audio playback to be cut off when the bot goes back to idle state.
Expected Behavior
The typing/recording indicator should remain active until the TTS audio is fully generated and sent to the user.
Steps to Reproduce
- Have TTS enabled
- Send a message that triggers a longer response
- The typing indicator stops while the audio is still being generated
- Audio gets cut off
Suggested Fix
Keep the typing controller active through the TTS generation phase in dispatch, only calling markDispatchIdle() after the audio message is actually sent.
Environment
- Clawdbot version: 2026.1.24-3
- Channel: Telegram
- streamMode: partial
Problem
When TTS is enabled, the typing indicator (chat action) stops as soon as the model run completes, but the TTS audio generation and sending happens after in the dispatch phase.
On Telegram (and likely other platforms), this causes the audio playback to be cut off when the bot goes back to idle state.
Expected Behavior
The typing/recording indicator should remain active until the TTS audio is fully generated and sent to the user.
Steps to Reproduce
Suggested Fix
Keep the typing controller active through the TTS generation phase in dispatch, only calling
markDispatchIdle()after the audio message is actually sent.Environment