Anyone having trouble with rime-ai repeating itsel...
# support
k
"role": "bot", "time": 1741978005302, "source": "", "endTime": 1741978012942, "message": "Great. Quick question. Are and your spouse typically more available mornings or evenings? Great. Quick question. Are you and your spouse typically more available mornings or evenings?", "duration": 7120, "secondsFromStart": 86.62 Call ID's: 546a3ef3-83df-4b6c-8e8e-4cba7081b6e1 It's the perfect voice for what I'm using and it seems like its only rime ai that does this
v
Yes, I've found that message repetition can happen with voice AI applications like Rime AI. Based on our documentation, there are several potential causes for this behavior: 1. **Network Issues**: Unstable connections can cause the AI to misinterpret or reprocess commands, leading to repetition in outputs. 2. **TTS Engine Bugs**: Bugs within the Text-to-Speech engine might cause unexpected behaviors like message repetition. 3. **Voice API Settings**: Misconfigurations in response parameters (like
numWordsToInterruptAssistant
and
maxDurationSeconds
) can cause incorrect processing. 4. **Echo and Feedback Issues**: If the AI's voice is being picked up by the microphone (e.g., when using loudspeaker), it might mistake its own output as new input. 5. **Dialogue Management Loops**: Unintended loops in the dialogue management system can trigger repeated outputs. 6. **Async Function Interference**: Improper use of asynchronous functions can result in repeated messages. For troubleshooting, you could try: - Checking your network stability - Reviewing your voice API configuration settings - Ensuring there are no feedback loops in your setup - Checking logs for error patterns Thanks for sharing the call ID - that will be helpful for tracking this issue. Rime AI does have advanced real-time voice synthesis capabilities, so it's important to ensure all settings are properly configured. Have you noticed any pattern to when the repetition occurs? Sources: - [Discord Thread about AI repetition issues](https://discord.com/channels/1211482211119796234/1211483291191083018/threads/1334109401463783434) - [Discord Thread about echo and feedback issues](https://discord.com/channels/1211482211119796234/1211483291191083018/threads/1337125226747662427) - [Discord Thread about dialogue management loops](https://discord.com/channels/1211482211119796234/1211483291191083018/threads/1333406713931632731)
k
Use gpt-4o and set the waitSecond to 0.6 and enable the endpointing as livekit. It will solve the issue.