### Answer Here are several ways to optimize speech-to-text accuracy in Retell AI: ## 1. Use Boosted Keywords Add commonly misrecognized words (names, brands, product terms) to the

boosted_keywords

setting. [(1) ]() This biases the transcriber model so those words are more likely to be transcribed correctly. [(2) ]()

javascript
boosted_keywords: ["retell", "kroger"]

[(1) ]() ## 2. Switch to "Accurate" Transcription Mode Set

stt_mode

to

accurate

instead of

fast

. The "optimize for accuracy" mode uses results with a higher endpointing setting, waiting longer with more context to generate more accurate transcripts (adds ~200ms latency) . [(3) ]() This is especially helpful for capturing entities like numbers and dates.

javascript
stt_mode: "accurate"

[(1) ]() ## 3. Enable Denoising Use the

denoising_mode

setting to filter out background noise:

javascript
denoising_mode: "noise-cancellation"

Or for more aggressive filtering:

javascript
denoising_mode: "noise-and-background-speech-cancellation"

## 4. Use Pronunciation Dictionary For words that are consistently mispronounced or misheard, add them to the pronunciation dictionary with IPA phonemes:

javascript
pronunciation_dictionary: [
  {
    "word": "actually",
    "alphabet": "ipa",
    "phoneme": "ˈæktʃuəli"
  }
]

## 5. Consider Vocab Specialization If you're in a specific domain like healthcare, set

vocab_specialization

to

medical

for better recognition of domain-specific terms (English agents only) :

javascript
vocab_specialization: "medical"

​ To learn more, see the sources I considered: [[1] Update Voice Agent](https://docs.retellai.com/api-references/update-agent) [[2] Get Voice Agent](https://docs.retellai.com/api-references/get-agent) [[3] Balance between transcription accuracy and latency](https://docs.retellai.com/build/transcription-mode) ​

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