### Answer I'm an AI assistant knowledgeable about Retell AI. The message you've shared appears to be a casual conversation suggestion rather than a question about Retell AI's features, documentation, or implementation. If you have questions about Retell AI—such as how to build voice agents, customize conversation flows, integrate APIs, configure prompts, or any other technical topics—I'd be happy to help using the official documentation and resources available to me. Let me know how I can assist you with Retell AI!

message has been deleted

Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.

### Answer To improve transcription accuracy for "haircut" with your British accent, here are several solutions: 1. Add Boosted Keywords [(1) ]() [(2) ]() Add "haircut" to your agent's boosted keywords list. This biases the transcription model to recognize specific words more accurately . You can configure this in the agent's global settings under Speech Settings [(1) ]() . 2. Set Language to en-GB [(1) ]() [(3) ]() Make sure your agent's language is set to

en-GB

(British English) rather than

en-US

. This affects speech recognition and pronunciation accuracy [(1) ]() . You can select this in the language dropdown in your agent's global settings . 3. Adjust Transcription Mode [(2) ]() Switch your transcription mode to be optimized for accuracy rather than speed . This helps prevent sentences from being cut off prematurely and improves overall transcription quality . 4. Configure Denoising If you're testing in a quiet environment and the system is missing short words, try setting denoising mode to "No Denoising" . This preserves more of the raw audio signal and can improve recognition of brief utterances . Implementation via API [(3) ]() If you're using the API, you can add boosted keywords like this:

json
{
  "boosted_keywords": ["haircut"]
}

These settings work together to improve ASR (Automatic Speech Recognition) accuracy [(4) ]() , which is critical for understanding what callers are saying .

​ To learn more, see the sources I considered: [[1] Configure global settings](https://docs.retellai.com/build/conversation-flow/global-setting) [[2] Increase transcription accuracy](https://docs.retellai.com/reliability/wrong-transcript) [[3] Update Voice Agent](https://docs.retellai.com/api-references/update-agent) [[4] Automatic Speech Recognition (ASR)](https://www.retellai.com/glossary/automatic-speech-recognition-asr) ​

Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.

### Answer I don't have specific documentation in my available sources that explains why phone call behavior would differ from web test call behavior in terms of the first sentence cutting off or script ordering changing. What I can note from the sources: web test calls and phone calls run through different infrastructure paths — phone calls involve the telephony layer (call routing, audio streaming, SIP connectivity) in addition to the core speech recognition, reasoning, and speech synthesis pipeline that both call types share [(1) ]() . This telephony layer is a separate point where issues like audio stream timing can surface that wouldn't appear in a browser-based test . I don't have documented details on your specific symptoms (first-sentence cutoff, reordered script) to give a confirmed root cause or fix. Check out these potentially useful sources for troubleshooting: [Documentation]() , [Nodes and edges]() , [Speech Processing]() Want me to give a general answer based on common causes, or do you have your call ID and agent setup details (e.g., which node type is used for the opening line) so I can look closer at what might be relevant?

​ To learn more, see the sources I considered: [[1] What It Takes to Build and Scale AI Voice Agents Effectively Without Them Breaking | Retell AI](https://www.retellai.com/blog/what-it-takes-to-build-and-scale-ai-voice-agents-effectively-without-them-breaking) ​

Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.