openclaw - 💡(How to fix) Fix [Bug] TTS sends duplicate text + audio, and format not configurable per channel [1 comments, 1 participants]

TL;DR

The issue can be resolved by modifying the hasAssistantVisibleReply function to correctly handle the audioAsVoice flag and by adding support for configurable TTS output formats per channel.

Guidance

• Review the hasAssistantVisibleReply function to ensure it correctly checks for the presence of text content when audioAsVoice is true, potentially by changing the condition to return resolveSendableOutboundReplyParts(params).hasContent && !params.audioAsVoice;
• Investigate adding a configuration option to the messages.tts section to allow specification of output format or path, or to enable per-channel format overrides
• Consider adding a check for the channel type (e.g., Feishu, WeChat) to determine the required audio format (e.g., opus/ogg for Feishu)
• Verify that the TTS provider (minimax) supports the required audio formats and can be configured accordingly

Example

function hasAssistantVisibleReply(params) {
    // Only consider text content if audioAsVoice is false
    return resolveSendableOutboundReplyParts(params).hasContent && !params.audioAsVoice;
}

Notes

The provided code snippet only addresses Issue 1, and a more comprehensive solution would require additional changes to support configurable TTS output formats per channel.

Recommendation

Apply a workaround by modifying the hasAssistantVisibleReply function as suggested, and consider adding support for configurable TTS output formats per channel in a future update, as this would require more significant changes to the codebase.