### Answer To improve transcription accuracy for names, emails, and street addresses, try these approaches: 1. Switch to "optimize for accuracy" transcription mode [(1) ]() Your current "speed" mode uses interim results with low endpointing, which can cause errors with entities like numbers, dates, names, and addresses . The "optimize for accuracy" mode waits longer with more context to generate more accurate transcripts, adding approximately 200ms latency but significantly improving entity capture . 2. Add Boosted Keywords [(2) ]() Add custom keywords to enhance the model's vocabulary for domain-specific terms . You can add up to 100 custom keywords . This is particularly helpful for specific names, street names, or terminology unique to your use case . 3. Adjust your custom STT configuration [(3) ]() If you're using custom STT settings, consider increasing your

endpointing_ms

value . Lower endpointing can cut people off early, causing parts of words to be missed [(4) ]() . 4. Check your language setting [(3) ]() Ensure your language setting matches the actual spoken language, as mismatched settings significantly increase Word Error Rate (WER) [(5) ]() . 5. Enable background noise handling if needed [(2) ]() [(5) ]() If background noise is an issue, adjust your denoising mode setting [(2) ]() . However, be aware that aggressive d

enoising can filter out short, low-energy responses . The combination of switching to "optimize for accuracy" mode and adding boosted keywords for commonly used names and addresses should provide the most immediate improvement for your specific issue with entity recognition.

​ To learn more, see the sources I considered: [[1] Balance between transcription accuracy and latency](https://docs.retellai.com/build/transcription-mode) [[2] Increase transcription accuracy](https://docs.retellai.com/reliability/wrong-transcript) [[3] Update Voice Agent](https://docs.retellai.com/api-references/update-agent) [[4] Discord message from @.boomz88 on 2026-04-16](https://discord.com/channels/1186424376300679209/1186462578310910112/1494348145805623418) [[5] Address Metric Issues](https://docs.retellai.com/ai-qa/address-metric-issues) ​

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